Front Cover
 Title Page
 Table of Contents
 Part one: Planning and on-farm...
 Part two: The plannning steps
 A final word on priorities
 Back Cover

Group Title: The planning stage of on-farm research : identifying factors for experimentation
Title: The planning stage of on-farm research
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00080101/00001
 Material Information
Title: The planning stage of on-farm research identifying factors for experimentation
Physical Description: vi, 85 p. : ill. (some col.) ; 26 cm.
Language: English
Creator: Tripp, Robert
Woolley, Jonathan
Publisher: CIMMYT
Place of Publication: Mexico D.F
Cali Colombia
Publication Date: 1989
Subject: Agriculture -- Research -- On-farm -- Planning   ( lcsh )
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Bibliography: Includes bibliographical references (p. 84-85).
Statement of Responsibility: Robert Tripp and Jonathan Woolley.
 Record Information
Bibliographic ID: UF00080101
Volume ID: VID00001
Source Institution: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 21373865
lccn - 90136603
isbn - 968612702X

Table of Contents
    Front Cover
        Page i
    Title Page
        Page ii
        Page iii
    Table of Contents
        Page iv
        Page v
        Page vi
        Page 1
        Page 2
    Part one: Planning and on-farm research
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
    Part two: The plannning steps
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
    A final word on priorities
        Page 83
        Page 84
        Page 85
    Back Cover
        Page 86
        Page 87
Full Text


T 4C-mrA E

The Planning Stage
Of On-Farm Research:

Identifying Factors
for Experimentation
Robert Tripp and Jonathan Woolley

The International Maize and Wheat Improvement Center (CIMMYT)
and the International Center for Tropical Agriculture (CIAT) are
members of a group of 13 nonprofit international agricultural research
and training centers supported by the Consultative Group on
International Agricultural Research (CGIAR), which is sponsored by
the Food and Agriculture Organization (FAO) of the United Nations,
the International Bank for Reconstruction and Development (World
Bank), and the United Nations Development Programme (UNDP).
Donors to the CGIAR system are a combined group of 40 donor
countries, international and regional organizations, and private

CIMMYT is an internationally funded, nonprofit scientific research and
training organization. Headquartered in Mexico, the Center is engaged
in a worldwide research program for maize, wheat, and triticale, with
emphasis on improving the productivity of agricultural resources in
developing countries. CIMMYT receives core support through the
CGIAR from a number of sources, including the international aid
agencies of Australia, Austria, Brazil, Canada, China, Denmark,
Federal Republic of Germany, Finland, France, India, Ireland, Italy,
Japan, Mexico, the Netherlands, Norway, the Philippines, Spain,
Switzerland, the United Kingdom, and the USA, and from the
European Economic Commission, Ford Foundation, Inter-American
Development Bank, OPEC Fund for International Development, UNDP,
and World Bank. CIMMYT also receives non-CGIAR extra-core support
from Belgium, the International Development Research Centre, the
Rockefeller Foundation, and many of the core donors listed above.

CIAT is a nonprofit agricultural research and training organization
devoted to the goal of increasing sustainable food production in
tropical developing regions. The core budget of CIAT is financed by a
number of donors. During 1988 these CIAT donors include the
countries of Belgium, Canada, China, France, the Federal Republic of
Germany, Italy, Japan, Mexico, the Netherlands, Norway, Spain,
Sweden, Swizerland, the United Kingdom, and the United States of
America. Organizations that are CIAT donors in 1988 include the
European Economic Community (EEC), the Ford Foundation, the
Inter-American Development Bank (IDB), the International Bank for
Reconstruction and Development (IBRD), the International
Development Research Centre (IDRC), the Rockefeller Foundation,
and the United Nations Development Programme (UNDP).

Information and conclusions reported herein do not necessarily reflect
the position of any of the aforementioned entities.

Responsibility for this publication rests solely with CIMMYT and CIAT.

Correct Citation: Tripp, R., and J. Woolley. 1989. The Planning
Stage of On-Farm Research: Identifying Factors for Experimentation.
Mexico, D.F. and Cali, Colombia: CIMMYT and CIAT.

ISBN 968-6127-02-X


v Preface
vi Acknowledgements
1 Introduction

3 Part One: Planning and On-Farm

3 A Review of On-Farm Research
5 Issues Related to the Planning Stage of OFR
7 The Use of Diagnostic Data in the
Planning Process
10 The Steps in the Planning Process
12 Summary

14 Part Two: The Planning Steps

14 Step 1. Identify Problems Limiting the
Productivity of the Farming System
26 Step 2. Rank the Problems
34 Step 3. Identify Causes of the Problems
46 Step 4. Analyze Interrelations among
Problems and Causes
52 Step 5. Identify Possible Solutions to the
55 Step 6. Evaluate Possible Solutions

75 Summarizing the Six Steps: Lists of

83 A Final Word on Priorities

84 Bibliography


This publication is a product of the on-farm research
training conducted by CIMMYT and CIAT over the
past several years. As our staff members have
developed and refined methods for teaching how to
plan and set priorities in on-farm research, they have
emphasized converting that experience into training

We are particularly pleased that this publication
represents a collaborative effort between two
international agricultural research centers. CIMMYT
and CIAT provide a variety of training in on-farm
research, and we believe it is worthwhile to compare
notes, learn from each other's experience, and, most
important, learn from the experience of national
program researchers and extension agents. One of
the things we consistently hear from our colleagues
in national programs is the necessity of coordinating
their efforts with ours in teaching on-farm research
methods. Publications such as this provide a forum
for establishing that collaboration. Work on different
crops under different conditions will necessarily lead
to some variation in procedures, but there is much
common ground from which general methods can be
developed. Those procedures can then be refined by
national program staff to fit their own requirements.

We believe that this publication will prove useful to
researchers and extension agents involved in
planning on-farm experimental programs. It should
also be helpful for national programs in developing
their own training activities. We anticipate that it will
promote more collaboration between international
research centers and national research programs in
the development of training methods and materials.
Suggestions for modifying and improving this
publication will be appreciated by the authors.

Donald L. Winkelmann John L. Nickel
Director General Director General


The methods outlined in this publication are the
product of the efforts of many people. Our first debt
is to the researchers and extension agents who have
participated in our courses and who studied, applied,
and improved the methods that we presented. The
methods have been tested at different stages of their
evolution with scientists from: Angola, Argentina,
Brazil, Colombia, Costa Rica, Ecuador, El Salvador,
Ethiopia, Guatemala, Haiti, Honduras, Indonesia,
Malawi, Mexico, Mozambique, Nicaragua, Pakistan,
Panama, Paraguay, Peru, and Rwanda.

This publication has gone through numerous drafts
and tests. At each step of the process, colleagues
from CIMMYT, CIAT, and other institutions were
kind enough to provide suggestions. We particularly
wish to thank Jorge Alonso Beltran, Derek Byerlee,
Mike Collinson, Cynthia Connolly, Willi Graf, Larry
Harrington, Alberic Hibon, Peter Hobbs, Roger
Kirkby, Ron Knapp, Marceliano Lopez Genes, Allan
Low, H.J.W. Mutsaers, Ed Pulver, Michael Read,
Gustavo Sain, Rigoberto Stewart, Armando Tasistro,
Alejandro Violic, Joachim Voss, Stephen
Waddington, Patrick Wall, and Michael Yates.

Additional thanks go to Anita Albert and Kelly
Cassaday, whose design and editing have made a
rather complicated document more accessible. The
skill of Maria Luisa Rodriguez and Beatriz Rojon, who
typed numerous drafts of this publication, is also
greatly appreciated.


On-farm research is a problem-oriented approach to
agricultural research that begins by diagnosing the
conditions, practices, and problems of particular
groups of farmers. Once the problems are identified,
a research program is designed to address them. A
key part of any such program is conducting
experiments on farmers' fields under farmers'
conditions and management. Those experiments are
then evaluated using criteria that are important to
farmers, and the results are used to make

This publication seeks to contribute to the
development of methods for on-farm research by
describing a method that can be used to decide
which factors have highest priority for experiments
that are planted in farmers' fields (the actual design
of experiments is not discussed). It also describes
how the planning of on-farm experiments can be
used to suggest other activities in support of on-farm
research. The procedures have been used in a wide
variety of settings and should be seen as flexible
guidelines, not rules.

An on-farm research program often generates so
much information and so many ideas that it is
difficult to decide which factors should be included
in on-farm experiments. Because these programs
usually have quite limited resources, and because
farmers prefer to consider recommendations that
address important problems, some decisions have to
be made in setting priorities for the experiments.
This publication presents six steps for discussing the
available evidence and deciding which factors should
be included in on-farm experiments.

The steps are a way to record and make explicit the
rationale behind the decisions that are taken in
selecting experimental factors. Furthermore, they can
serve as an agenda for a meeting to plan on-farm
experiments. The participants in the meeting would
be the researchers and extension personnel involved
in diagnostic and experimental activities. Other
persons, such as specialists in subjects that are
particularly relevant to the research program, may
participate as well. The meeting should take place
sufficiently in advance of planting so that
researchers can arrange materials and sites for the

experiments. The planning may take several days,
including the identification of factors and
experimental design. It may take place in an office or
a conference room, and the facilities should be
adequate for the examination of data, the
interchange of ideas, and the debate and
compromise that characterize planning. These
planning steps can be used before each cycle of
experimentation; indeed, they can be used at any
time for reviewing the rationale of an on-farm
experimental program.

The primary audience for which this publication is
intended comprises researchers and extension agents
involved in on-farm research. To guide readers
through the planning method, a detailed description
of each step in the process presents its objectives,
the activities involved, and its relation to the other
steps. The descriptive sections are complemented by
a comprehensive example that uses data from one
hypothetical research area and is carried through all
of the steps (see box, p. 21). The example is
purposely complex to illustrate various issues that
might arise in the course of planning.

Aside from its use as an aid to researchers and
extensionists, this publication is also designed to be
used in training courses that address planning
methods. For that purpose it is best if course
participants have access to data from an on-farm
research program, preferably data they have
developed themselves. After each step is introduced,
the participants can use their data to work through
it, and their conclusions can be discussed before
proceeding to the next step.

This document is divided into two parts. The first
provides a brief review of on-farm research and
discusses some features that are particularly relevant
to planning. The second part presents the six steps
that constitute the planning method and the lists
used to summarize the conclusions.

Part One:

Planning and

On-Farm Research

A Review of On-Farm Research

On-farm research (OFR) is a set of procedures for
adaptive research whose purpose is to develop
recommendations for representative groups of
farmers. In OFR, farmers participate in identifying
priorities, managing experimentation, and evaluating
results. Procedures for OFR can be divided into five
stages, among which there is considerable overlap
and feedback.

1) Diagnosis
The diagnostic stage involves collecting and
analyzing information in order to design on-farm
experiments. Diagnostic activities may include a
review of secondary data, interviews with local
officials, informal farm surveys consisting of farmer
interviews and field observations, and formal surveys
with a questionnaire. The purpose of the initial
diagnostic activities is to gather enough information
to describe the basic features of the research area,
to identify problems that limit farmers' productivity,
and to begin considering possible improvements in
farmers' practices. The information obtained from
diagnostic activities can be used to design the first
cycle of on-farm experiments. Of course, diagnosis
does not end once the first experiments are planned.
Many of the experiments themselves are designed
for diagnostic purposes, and during the
experimentation stage the need often arises for
further diagnostic activities, including informal
observations and formal studies.

2) Planning
The planning stage of OFR is the focus of this
publication, which describes six steps used to
identify experimental factors to be included in on-
farm experiments, as well as to suggest other
research activities. During the first year of work,
information generated during the diagnostic stage is
used to design a set of on-farm experiments. In
subsequent years, data from those experiments play
an increasingly important role in planning. The six
steps outlined here are as useful for the first year,
when only survey data may be available, as for
subsequent years, when results of on-farm
experiments are available to plan future work. But
though they help identify factors for

experimentation, the six steps are only part of the
process of experimental design. The other aspects of
that process are treated in detail in another
document. 1

3) Experimentation
On-farm experiments are planted in the fields of
representative farmers and examine a small number
of experimental variables. Those experiments may be
described and classified in a number of ways, but
regardless of classification most of them progress
from exploring production problems, to testing
possible solutions, and then to verifying
recommendations and demonstrating them with
farmers. In this document, some factors will be
referred to as "exploratory," which means that they
seek to better define and characterize a particular
production problem. Exploratory factors therefore
serve a diagnostic function, and their results
contribute to the planning of the next cycle of
experiments. Other factors will correspond to
possible solutions for production problems that are
already well understood. The experimental variables
in an on-farm experiment may represent exploratory
factors, possible solutions, or some combination of
the two. Nonexperimental variables are usually set at
the level of representative farmer management.2

4) Assessment
The results of the on-farm experiments should be
analyzed carefully. The analysis requires an
assessment of farmers' reactions and opinions, a
thorough agronomic interpretation, and careful
statistical and economic analysis. The results of the
assessment are then used to plan future research
and to make recommendations for farmers.

5) Recommendation and Diffusion
When researchers are confident that they have
enough information, they can formulate
recommendations. In a system of on-farm research

1 The organization of factors in experiments, taking into
account the resources available, is presented in Woolley
2 See Woolley (1987) for a discussion of the exceptions.

that functions well, extension agents participate in
the entire process and so are able to transfer
recommendations to farmers with skill and
confidence. When farmers are actively involved in
the research process, they participate effectively in
the diffusion of new technologies. By monitoring
farmers' opinions and use of new technologies,
researchers can improve their understanding of
farmers' needs and preferences.

Issues Related
to the Planning Stage of OFR
There are well-established procedures for most
stages of on-farm research. Survey techniques, as
well as methods for statistical and economic
analyses, are described in several publications.3 It is
not so easy to find references on the planning stage,
even though it may be the most critical part of the
process.4 During the planning stage researchers
become committed to a set of on-farm experiments
and other activities that will absorb an important part
of the research budget. It is essential that the
rationale for these experiments be well conceived
and carefully documented.

The method presented here concentrates on
designing on-farm experiments intended to develop
technologies that can improve farmers' productivity
in the short term (three to five years). This type of
adaptive research relies heavily on technologies such
as new crop varieties or management practices
which have been developed through applied research
on and off the experiment station. However, longer
term priorities for technology development can be
suggested by information generated during OFR, and
the steps described here can be used to propose
new themes to be addressed by the research

3 A description of survey techniques for OFR is found in
Byerlee et al. (1980). Statistical analysis is described in any
number of books, including Gomez and Gomez (1984).
Economic analysis for OFR is presented in CIMMYT (1988).
4 There are some exceptions. Valuable advice on planning in
adaptive agricultural research can be found in Mutsaers
(1985), Van Der Veen (1984), and Huxley and Wood (n.d.).

On-farm research is also dependent on agricultural
development policy, which often directs research
toward particular regions, crops, or types of farmer.
But OFR may also contribute to the formation of
policy. This planning method can be used to identify
suitable issues to be pursued with officials
responsible for implementing policy or for extension.

It is assumed that the steps described here will be
used to plan a research program in one specific area
of a country and that the area's boundaries have
been defined before planning begins. In addition, it is
assumed that the research will focus on one or a
few crop enterprises, considered individually or as
part of a cropping pattern.5 Which enterprises)
researchers study may be determined in advance by
the research program or identified in the course of
diagnosis and planning. In any case, limited research
resources make it necessary to concentrate on a few
enterprises at a time.

Planning an experimental program also requires an
understanding of how farmers adopt innovations.
Although in certain situations farmers may need to
make simultaneous changes in a number of their
practices, it is always advisable to consider offering
farmers intermediate steps toward the adoption of a
package of practices. Most farmers are very cautious
and tend to adopt one or a few new inputs or
techniques at a time. This stepwise adoption
behavior has important implications for planning an
experimental program, because it may influence the
number of factors that are tested at any one time
and the order in which they are tested.

5 This paper deals only with crop research. There is every
reason to believe that most of the principles described here
are applicable to animal production research as well, but we
do not have enough experience to speak with authority on
that subject.

The Use of Diagnostic Data
in the Planning Process

Effective planning depends on the information
developed during the initial diagnostic activities, the
diagnosis that takes place afterward (supplementary
surveys, observations, crop or soil samples, etc.),
and the results of the experiments themselves. A
diagnosis has four goals:

To describe the circumstances and practices of
representative farmers;

To identify problems limiting the productivity of
the resources available to farmers;

To understand the causes of those problems; and

To begin to consider possible solutions.

Those goals can rarely be attained completely during
an initial diagnosis. Even after several years of
experiments questions remain and new ones arise.
Procedures for planning OFR must be designed to
allow researchers to take account of uncertainties
and to suggest the most efficient way to resolve

A description of farmers' circumstances should
encompass both natural circumstances such as soils
and climate and socioeconomic circumstances such
as farmers' resources, local institutions, and
markets. It is attained through a review of secondary
data and various kinds of surveys, which also
provide a description of production practices in major
crop enterprises. Understanding how farmers'
practices are conditioned by their circumstances is
an important part of the diagnosis.

Besides being descriptive, the diagnosis should be
analytic. Beginning with the diagnostic stage and
extending into planning and experimentation,
researchers should think in terms of problems,
causes, and solutions. These three terms are central
to the discussion of planning as it is presented in
this publication because they provide a way to order
the establishment of research priorities and
correspond to the questions, "What's wrong?",
"Why?", and "What can we do?".

What is meant by the term "problem"? The
objective of OFR is to identify cases of low
productivity and try to resolve them. Low produc-
tivity is reflected by low yields and incomes or by
high production costs. But simply identifying an
instance of low yields or high production costs will
not provide direction to a research program.
Problems must be described in greater detail.

In many instances problems can be described as
biological limiting factors, such as nitrogen deficiency
or weed competition. Other problems involve
resource use. It may be that inputs are not used
efficiently, that land or labor could be employed
more intensively or production costs could be
lowered, or that a crop of higher value might be
substituted for the current crop. In any of these
cases, the fact that productivity could be improved
is evidence of a problem, which should be described
either in terms of a biological limiting factor or
inefficient resource use.

The careful specification of a problem is essential but
frequently is not enough to indicate a course of
action. The causes of a problem must be identified
as well, because knowing the causes helps to
determine possible solutions. For example, suppose
there is a disease that affects maize that is planted
late. The maize is planted late because farmers plant
their sorghum first. The problem is the maize
disease. One of its causes is late planting, which in
turn is caused by the need to plant the sorghum on
time. In this case, simply proposing an earlier
planting date for maize would not be a possible
solution, because it would interfere with planting
sorghum. Possible solutions that are suggested by
analyzing the causes of the problem include not only
a disease-resistant maize variety but also an earlier
maturing maize variety or a quicker method for
planting sorghum.

It is important to maintain a clear distinction
between problems, causes, and solutions, although
that is not always easy. In specifying problems
researchers will naturally think ahead to possible
solutions. Through their own experience and
knowledge of the literature, researchers will be
aware of crop varieties, products, and techniques
and will inevitably compare farmers' practices to the
technologies that are available. But in the planning
process the first step is a careful specification of
problems. That step is followed by an identification
of their causes, and only after that are possible
solutions considered.

For this planning method to be most effective, the
diagnostic survey and any other diagnostic studies
preceding the first experiments should be organized
around the concepts of problem, cause, and solution.
Surveys should first describe farmers' conditions and
practices and then move on to an initial identification
of problems. As the survey work proceeds,
researchers must try to identify the causes of those
problems and, where appropriate, explore the
feasibility of possible solutions. Those latter
objectives may be pursued by conducting more
detailed informal interviews and field observations,
and/or designing a questionnaire. Researchers should
be familiar with the steps in the planning process
before beginning diagnostic work.

This method assumes that the research is aimed at
well-defined groups of farmers.6 A careful definition
of the types of farmers who are likely to benefit
from the research is essential to the planning
process. That definition is important at two points:
first, when clearly specifying the characteristics of
farmers who share a particular problem, and second,
when identifying farmers in that group who are
eligible for a particular solution.

The Steps in the Planning Process

There is certainly no single correct approach to
planning. Many methods have been developed for
planning in administration, business, and other
fields.7 This method of selecting factors for on-farm
experiments is derived from considerable experience,
especially in conducting practical training courses in
a wide variety of settings. But the method should be
taken as a guide rather than a rigid set of rules, and
researchers may want to modify it to suit their own

The method is based on a series of steps (Figure 1)
corresponding to the distinctions between problems,
causes, and solutions. Step 1 is the identification of
problems and, where necessary, the specification of
means for gathering further information to define a
problem. Once the problems are listed, Step 2
subjects them to a review in which a rough order of
priority is assigned to each problem according to the
number of farmers affected, the importance of the
crop, and the seriousness of the problem.

6 One term for describing a group of farmers with similar
circumstances and for whom a particular recommendation is
appropriate is "recommendation domain." The use of
recommendation domains is discussed in greater detail in
Harrington and Tripp (1984). Alternative terms for describing
these groups of farmers are used in the literature.
7 For a summary of some planning methods see Delp et al.

Figure 1.
Steps in the planning process

Further evidence
required to identify
or evaluate problems

Analyze interrelations
among problems
and causes

Further evidence
required to determine
causes of problems

List A List B
Factors for Other
experimentation diagnostic

List C
Longer term

List D

Step 3 involves identifying causes for each of the
problems and, where the causes are not known,
specifying the evidence required for their
identification. Step 4 summarizes the interrelations
between problems and their causes.

In Step 5 researchers will seek possible solutions to
each problem for which there is sufficient evidence.
The solutions must take into account what is known
about the causes of the problems and should include
several alternatives, if possible. Step 6 then narrows
the list of possible solutions by evaluating each one
for potential benefits, ease of adoption by farmers,
and ease of investigation.

The results of the six steps are summarized in four
lists (Figure 1). The first presents factors for on-farm
experiments (either exploratory factors that help
examine a problem or its causes, or factors that
correspond to possible solutions). The second list
contains suggestions for other diagnostic activities to
help understand particular problems or causes. A
third contains suggestions for research activities, on
farms or the experiment station, that respond to
longer term research needs. The fourth list suggests
opportunities to transmit the data of on-farm
research to officials responsible for policy
implementation or extension.


On-farm research can be viewed as having five
stages: 1) diagnosing farmers' conditions and
problems; 2) using that information to plan a
program of experiments; 3) carrying out a program
of on-farm experiments; 4) analyzing the results; and
5) deriving recommendations for farmers. This
document focuses on the second stage, planning.

Several characteristics of on-farm research are
relevant to the planning stage. First, the research
should focus on a small number of crop enterprises.
The interactions of these enterprises with the rest of
the farming system must be taken into consideration,
but it will not be possible to do research on many
enterprises at the same time. Second, the research
program should focus on the possibility of making
incremental changes in farmers' practices, beginning

with farmers' current practices and planning a
strategy that allows them to gradually improve the
productivity of their resources. Third, the planning
should be based on a diagnosis of the research area
that includes a description of farmers' conditions and
practices, an identification of production problems,
and an explanation of their causes.

The planning process described here is divided into
six steps. The first is an identification of problems
and a review of supporting evidence. The second is
a ranking of these problems in order of importance.
The third step is an identification of the causes of
the problems and a review of supporting evidence.
The fourth is a consideration of the interrelationships
among all of the problems and causes that have
been identified. The fifth step is a list of possible
solutions to the problems. The sixth step is an
evaluation of the suggested solutions and a selection
of those solutions that seem to have the best
chances of success.

The products of these six steps are summarized in
four lists:

A) Factors for on-farm experiments (these factors
address problems and suggest solutions that
researchers feel can have an impact on the
farming system in the short term, within five

B) Other diagnostic activities such as surveys,
observations, and laboratory tests to identify
problems or their causes;

C) Longer-term research activities on the
experiment station or through on-farm trials, to
develop solutions that require more time; and

D) Suggestions for developing support from
institutions responsible for extension, credit, or
input policy in order to ensure that the results of
the research program can be used by farmers.

Part Two:

The Planning Steps

Step 1.
Identify Problems Limiting the
Productivity of the Farming System

On-farm research is a problem-oriented approach to
adaptive research, in which priorities are set
according to an understanding of the production
problems of particular groups of farmers. It is thus
very important that, as the first step in the planning
process, problems be carefully defined.

For the purpose of this publication, "problems" are
biological limiting factors or inefficiencies in the use
of resources that restrict the productivity of a
farming system. Problems should be described in a
way that clearly illustrates their relationship to low
yields, low income, or inefficient resource use.

The type of evidence used in identifying a problem
should also be noted. Was the problem identified
through experiments, a survey, field observations,
laboratory tests, or other methods? In some cases
where there is not yet sufficient evidence to confirm
that the problem exists, researchers will list tentative
problems. In those cases it is necessary to describe
what additional evidence (from experiments or other
methods) is required to further specify or confirm
the problem.

Problems should always be identified jointly by
researchers8 and farmers, and researchers should
always take account of farmers' perceptions and
priorities. One of the principal goals of any diagnosis
is to understand what farmers see as their problems.
Although farmers will not always be aware of some
problems that researchers are able to detect, they
will often point out factors that researchers might
otherwise have missed.

In addition, the problems that researchers identify
should always be seen in the context of the entire
farming system. For example, it may appear that
interplant competition caused by high plant
population is a problem in maize fields. But farmers

8 In this publication, on-farm "researchers" may be
researchers, extension agents, or both working as a team.

may be managing their maize to provide both grain
and forage. Once the importance of animals to the
farming system is understood, high plant populations
may cease to be seen as a problem.

The resources available to farmers are important
elements of the farming system and help define
which problems farmers face. Those resources
should be taken into account when considering how
to improve the productivity of farming systems. In
some situations where land is abundant and labor is
scarce, farmers may find that planting relatively large
areas, managing them less intensively, and getting
lower yields than they might otherwise obtain from a
small, well-managed plot is a rational use of their
resources. Thus the definition of problems requires a
good understanding of the farming system, an
appreciation of farmers' resources, perceptions, and
priorities, and a continual dialogue between farmers
and researchers.9

The type of research described in this document
focuses on one or a few crop enterprises, and the
importance of choosing those crops in the context of
the farming system must be emphasized. When the
need for subsistence food is a priority, crops
essential to the local diet may receive attention first.
In other cases cash income may be most important,
and crops that are marketed should receive priority.
In still other cases, animal production or the
maintenance of draft animals is crucial to the
farming system, and work on forage crops is most

Considerable care should be devoted to defining
problems because this step will determine the course
of subsequent steps in the planning process. It may
be useful to divide the following discussion between
two general types of problems, those related to
biological limiting factors and those related to
inefficient resource use, with the understanding that
these categories overlap to some extent. After that
discussion, attention will turn to the kind of evidence
needed to consider a problem in subsequent planning

9 An excellent example of this type of dialogue is found in
Rhoades and Booth (1982).

Limiting Factors
Those things that agronomists normally think of as
limiting factors, such as nutrient deficiencies, too
much or too little moisture, weeds, or insects and
other pests (including storage pests), are probably
the most common examples of problems, as the
term is defined here. Limiting factors that exhibit
year-to-year variability, such as drought or frost, are
sources of risk for farmers and must be considered,

Limiting factors need to be described as precisely
as possible:

* The particular nutrient deficiency needs to be
specified: The bean crop is affected by
phosphorus deficiency.

* The type of insect responsible for crop damage
needs to be identified: The maize suffers from
stem borer attack approximately one year in

* The period of the crop cycle when drought is
most likely to occur needs to be established:
Drought often affects the wheat crop late in the
growing season.

Distinguishing problems from causes and solutions-
In defining problems equated with limiting factors, it
is important to distinguish them from causes and
solutions. For instance, if farmers lose wheat to late-
season drought when the crop is planted late, the
problem is "late-season drought," not "late
planting," which is one of the causes. Similarly, a
problem might be described as "severe stem borer
attack in maize" but not as "farmers don't use
insecticides," which expresses a possible solution
(see also box, p. 17).


Symptoms and problems-To identify limiting
factors, agronomists use a wide variety of evidence,
including symptoms of nutrient deficiencies,
diseases, or pests; abnormal growth characteristics;
or the analysis of yield components. The emphasis
here is not on the skills required to interpret that
evidence, but rather on the necessity of describing
problems as precisely as possible. In developing
evidence for a problem it is important to distinguish
between the symptoms and the problem itself. If a
symptom provides clear evidence of a problem, then
there is no difficulty. For example, if striped leaves
on maize plants, combined with other evidence (soil
or tissue analyses, or exploratory experiments), point
clearly to a magnesium deficiency, then the
magnesium deficiency should be taken as the
problem. But striped leaves may indicate one of
several mineral deficiencies. If it is only suspected
that this symptom is caused by magnesium
deficiency, then that deficiency should be listed as a
tentative problem, and additional evidence should be
sought. Striped leaves themselves should not be
listed as a problem.

Similarly, such abnormal growth characteristics as
short internodes in beans or barren plants in maize
certainly indicate problems but they do not provide
enough information for researchers to think about
solutions. More evidence must be obtained before
the problem can be defined. For example, barren
maize plants might be associated with a nutrient
deficiency or interplant competition. One or more of
the factors most likely to be associated with the
abnormal growth characteristic should be considered
as tentative problems, and more evidence should be
gathered before the problem is finally defined. (Ways
to develop evidence for tentative problems are
described on p. 24.)

Interactions between problems-There may be
interactions between two or more of the problems
that have been identified. If both nitrogen deficiency
and weed competition are problems, the nitrogen
deficiency may partly result from the weed
competition. Since the possibility of such interactions
is dealt with in Step 3, the two problems should be
listed separately in this first step.

Inefficient Resource Use
Problems that limit the productivity of the farming
system may also be related to inefficient resource
use. Such problems are often recognized because
researchers are aware of more efficient alternative
practices, but the problems should be defined in
such a way that various solutions can be considered.

Sometimes there may be evidence of inefficient use
of inputs. Problems of this nature include, among
other examples, the excessive use of pesticides, the
use of inappropriate products (compound fertilizers
where single-nutrient fertilizers are sufficient), or the
misuse of irrigation water. High costs of particular
operations, such as weeding or tillage, are problems
that may also arise from inefficiency.

Example: Farmers apply a basal dressing of
10-30-10 fertilizer, but there is no evidence of
a response to phosphorus or potassium.

Perhaps in some cases land or labor could be used
more effectively. Land may be idle, although labor is
available; opportunities for relay cropping or
intercropping may exist; it may be possible to exploit
a second growing season. Those sorts of situations
may become apparent through an analysis of
resource use, or might be suggested by such
occurrences as seasonal food or fodder shortages.

Example: Farmers leave most of their land idle
during the minor rainy season.

Farmers' incomes might be improved by changing a
crop or crop variety. If one type of bean receives a
higher price in the market, or if a new crop offers
possibilities for raising farmers' incomes, then those
options should be examined. However, caution
should be exercised when considering alternative
varieties or crops until there is assurance that
markets can accommodate the new product.

Example: Farmers receive low prices for their
maize crop, and there is an increasing demand
from nearby towns for vegetables.

Institutions and Infrastructure
Factors related to institutions and infrastructure are
often mentioned as problems but do not really

qualify. Diagnosis in on-farm research often reveals
institutional inadequacies--poorly developed markets,
low crop prices, lack of extension or credit, bad
roads, and so forth. Such factors are undeniably
important elements of the farmers' environment and
should be included in a description of the research
area, but they do not qualify as problems in the
context of planning an on-farm experimental
program. In many cases, however, inadequate
institutions or poor infrastructure may be considered
causes. If a fertility problem is partially due to the
lack of credit for purchasing fertilizer, or if extension
advice is not available for helping farmers control
insects, then on-farm research data from
experiments and surveys can be brought to the
attention of policymakers who deal with credit or
extension. That kind of information is included in List
D, so that it can be reviewed by researchers and
decisions can be taken about the most effective way
to discuss those issues with policymakers.

Presenting Problems to be Addressed in Planning
All of the problems that have been identified should
be listed. Table 1 (p. 22) shows one way to do so. It
lists all of the problems identified in the example
research area (see box, p. 21). This example will be
followed throughout the rest of the publication.
Table 1 also lists the evidence available to confirm
each problem and indicates where additional
evidence is required.

Evidence of problems-In addition to identifying the
problems, researchers should specify the type of
evidence available to confirm or support the
existence of each. Much of the evidence for
problems may come from the results of on-farm
experiments. Exploratory experiments give evidence
of responses to various factors and interactions
among them, as is the case with nitrogen and
phosphorus deficiencies in maize (Table 1, column 2,
1, 2). Further experiments help researchers refine the
definition of problems and identify new ones, as
when experiments with earlier maturing maize
varieties pointed the way to identifying drought as a
problem (3). Field observations may provide evidence
in other cases, as with anthracnose in beans (6), and
farmer surveys can reveal additional information, as
with the high cost of weeding in maize (4).


Table 1
List problems (Step 1)

Nitrogen deficiency
in maize

deficiency in maize

Drought stress in
maize at ear filling

High cost of
weeding maize

Nitrogen deficiency
in beans

"Poor soils" mentioned
by farmers; two years of
experiments have shown
response to nitrogen in

"Poor soils" mentioned
by farmers; two years of
experiments have shown
small, noneconomic
response to phosphorus.
Many fields show usual
signs of phosphorus

Field observations.
Two years of
experiments have shown
significant yield
advantage for earlier
maturing variety.

Survey data show that
farmers do two and
sometimes three hand
weedings. The cost of
labor in the study area
has risen 50% in the last
three years.

"Poor soils" mentioned
by farmers; yellow
leaves suggest nitrogen

No more evidence

No more evidence

There is sufficient
evidence that problem is
important; examine
meteorological data to
determine frequency of
drought and whether the
last two years were
representative (List B).

There is sufficient
evidence that problem is
important; interview
farmers who are
beginning to use
herbicides regarding their
experiences and opinions
(List B).

No more evidence

Anthracnose attack
on bean pods
(about one year in

Root rots of beans
during crop
identified as a

Low plant
population in beans

Broad-leaf weed
competition in
beans (tentatively
identified as a

Waterlogging in
bean fields

Progen deficiency
in tobacco

Field observations,
conversations with
extension agents, and
interviews with farmers.

A few farmers' reports
during interviews (other
farmers probably have
beans that suffer from
this problem but it goes

Field observations (not
certain to what extent
the problem is due to
root rots, poor seedbed,
formation of soil crust,
or poor seed quality).

Field observations
indicate that there may
be some yield loss:
farmers complain of
weeds in beans.

Field observations.

Symptoms of nitrogen
deficiency are apparent

No more evidence

Field sampling to
determine severity and
laboratory tests to
confirm species
responsible (List B).

No more evidence
required regarding the
problem, but more
evidence of its causes)
is needed-see Step 3.

experimentation is
needed to see if there is
a response to broad-leaf
weed control (List A).

No more evidence

No more evidence

Each year, a thorough review of previous
experimental results and data from other research
activities is required, as well as a new listing of
problems, in accordance with the evidence available.
In this process some problems may be eliminated
from the original list and new ones added.

Tentative problems and further evidence-Often
researchers will note problems that require more
evidence. If little evidence is available regarding a
problem, then considering solutions is rarely
worthwhile. Instead, more effort should be spent in
confirming the existence or the nature of the
problem. The degree of confidence that researchers
have in the evidence for a specific problem will
determine how they should proceed. There are
several possibilities.

In some cases, researchers may feel that there is
enough evidence to justify examining solutions to a
problem but may still suggest gathering additional
evidence. To better quantify the problem of drought
in maize (3), researchers propose an examination of
meteorological data. To further investigate the high
cost of weeding maize (4), they will interview
farmers to gather additional evidence regarding the
costs of different weeding methods.

Sometimes a problem is only tentatively identified,
and more evidence is required before solutions can
be sought. Broad-leaf weed competition in beans (9)
is one such case, and an exploratory experiment is
proposed to judge its seriousness. The experimental
factors) are chosen to obtain a clear "yes" or "no"
answer to the question: Is broad-leaf weed
competition a problem in bean production in the
research area?

In cases such as this the experimental factors) may
be possible solutions to the problem (such as a
particular herbicide). But it must be realized that here
the "solutions" really serve a diagnostic function,
and that once the problem is confirmed other
solutions (such as extra hand weeding) may be
considered. In other cases, experimental factors may
be used for diagnosis although they would not be
feasible solutions. For example, soil liming might be
used to determine how much soil acidity reduces

yields even if liming is too expensive to be a
potential solution for farmers. Instead, if the problem
proves to be important researchers might look for a
variety tolerant to acid soils.

Finally, there are instances in which nonexperimental
evidence may be useful to help identify or
characterize a problem. In the case of root rots in
beans (7), field sampling and laboratory tests are
proposed to confirm the existence of the problem.

If researchers believe that no additional evidence for
a problem is required, then they can proceed to
examine causes. If additional evidence is sought, the
type of evidence required should be noted. When so
little evidence exists that researchers are uncertain if
the problem is really present, then they should list
the type of evidence required. That evidence may be
an exploratory experimental factor (to be included in
List A), or some other type of diagnostic information
(List B).


The first step in planning on-farm experiments is to
list problems that limit the productivity of the
farming system under study. These problems may be
described as either biological limiting factors or as
inefficiencies in resource use. Care should be taken
to describe the problems as precisely as possible.

Evidence for each problem should also be presented.
It may be derived from previous experiments,
surveys, or other diagnostic techniques. Researchers
should decide if they have enough confidence in the
evidence to identify or confirm the problem.

When the problems and supporting evidence have
been listed, they are ready for consideration in
Step 2, where they will be ranked in rough order
of importance.

Step Two.
Rank the Problems

Although it may be fairly easy to describe many
problems encountered in a given research area, it is
usually impossible to investigate more than a few at
a time. Research programs have limited budgets and
priorities must be set. Furthermore, the idea of
investigating a few priority issues at one time is
consistent with the strategy of on-farm research,
which is to make gradual changes in farming
systems. It is therefore important to evaluate
problems that may become research topics to
determine which should receive priority.

Step 2 is only an initial ranking of the problems listed
in Step 1. If the list of problems is very long (greater
than 10 or so), it may be possible to eliminate some.
But even if no problems are eliminated, an initial
ranking helps establish a sense of priorities. Problems
that receive a low ranking but which have important
relations to other problems (see Step 4) or which
have easily accessible solutions (see Step 6) may
certainly be addressed in the experimental program
being planned. Problems that receive low rankings in
this review and lack either obvious interactions with
other problems or easy solutions may be eliminated
from consideration altogether.

Recall that some problems considered in Step 1 may
not be well defined. Even so, they should be
reviewed. A tentative problem may still be
sufficiently important to warrant investigation either
through experimentation or other techniques for
gathering data.

Problems should be ranked every year. The ranking
assigned previously should be reviewed in light of
new evidence from experiments and other sources.
Over time the importance of certain problems may
diminish, whereas others may be assigned higher
priority for future work. Maintaining consistency in
the experimental program is an important
consideration when researchers rank problems.
Priorities may change from one year to the next,
depending on the results of the research program,
but the content of the on-farm experiments should
exhibit a logical progression from year to year rather
than skip from one topic to another.

There are numerous ways of assigning priorities to a
set of research problems. The method suggested
here employs three criteria to do so:

The distribution of the problem;
The importance of the particular crop enterprise to
the farming system; and
The loss of yield or income for which the problem
is responsible.

The application of these criteria to the example is
presented in Table 2 (pp. 28-29).

Distribution of the Problem
It is necessary to specify which farmers are affected
by the problem. How many farmers in the research
area grow the crop (or crops) in question? Of those
farmers, how many have crops affected by the
problem? Finding the answers to those questions
may require only a straightforward estimate of the
proportion of farmers in the area who grow the crop.
All tobacco farmers seem to have nitrogen deficient
crops (problem 11, Table 2), but few farmers grow
tobacco and the problem gets a very low rating.10

A rough estimate of the number of farmers who
grow the crop and have the problem is necessary. If
only certain farmers seem to have the problem, a
description of that group should be made. Most
maize farmers have the problem of nitrogen
deficiency, so it gets a high ranking. On the other
hand, drought is primarily a problem for maize
farmers in the north, so its rating is lower. Few
farmers have waterlogging in their beans (10), so it
gets a low ranking. The problem of root rots in
beans is still unconfirmed and only tentatively
identified as corresponding to a particular group
of farmers.

Importance of the Crop Enterprise
In some cases a target crop or crops will already
have been selected for on-farm research because it
is either included in the mandate of the research
organization or in national agricultural development

10 Specifying which farmers are affected by a particular problem
may also involve weighting the estimate in favor of particular
kinds of farmers. For example, government policy may give
higher priority to a problem affecting a crop grown by small-
scale farmers than to a problem found in a crop produced by
larger scale farmers.

Table 2
Rank problems (Step 2)
(XX = very important; X = somewhat important; 0 = not important)

Nitrogen deficiency
in maize

deficiency in maize

Drought stress in
maize at ear filling

High cost of
weeding maize

Nitrogen deficiency
in beans

Anthracnose attack
on bean pods
(about one year in

Most farmers.

Most farmers.

Farmers who live in
the northern part of
the research area,
which is more prone
to drought.

Most farmers.

Most farmers.

Most farmers.







Root rots of beans
during crop
establishment (?)

Low plant
population in beans

Broad-leaf weed
competition in
beans (?)

Waterlogging in
low-lying bean

Nitrogen deficiency
in tobacco

About half of the
farmers (?). (Those
who plant beans
every year on the
same field are
probably affected

Most farmers.

Most farmers (?).

A few farmers who
have low-lying

Farmers who grow
tobacco (less than
10% of all farmers).

* Tentative problem; more evidence required












policy. But in other cases research topics must be
selected from among the problems affecting various
crops. To do so, researchers must determine
whether the crop enterprise is a significant source of
income or subsistence for the farmers who grow it,
and/or if it utilizes significant amounts of farmers'
land, labor, or capital. In other words, if changes
were made in the productivity of the resources
devoted to the enterprise, how much of a
contribution would they make to the system's overall
productivity? Sometimes a problem extends to more
than one crop (e.g., a problem related to effective
tillage) and merits a higher ranking than is given to a
problem affecting just one crop. At other times the
problem relates to a new or recently introduced crop,
and researchers may want to cautiously estimate the
crop's potential importance.

In the research area, maize occupies more land than
beans and is the principal item in the diet. It has the
highest ranking. Beans, which are important in the
diet and increasingly important as a source of cash
income, receive a medium ranking. Tobacco also
gets a medium ranking, because although few
farmers grow this crop, it makes a fair contribution
to the incomes of those who do. (Remember, the
number of farmers growing each crop was
accounted for when the distribution of the problem
was determined.) Most farmers grow a few
tomatoes in their gardens, but tomatoes are
unimportant in local incomes or diets, so problems in
tomato production were not considered.

Seriousness of the Problem
Researchers should estimate whether the problem is
responsible for a significant yield loss or serious
inefficiency in resource use. That judgment may be
difficult, especially if a problem is not well defined,
but an estimate of a problem's potential importance
should be attempted. In making the estimate, two
elements of the problem are considered:

* The severity of the loss. For farmers whose crops
are affected by the problem, how much yield per
hectare is lost because of it, or how much income
per hectare is lost because of inefficient resource

* The frequency of the problem. Does the problem
occur every year, or only in a certain percentage
of years?

The product of these two elements (severity times
frequency) gives an estimate of the seriousness of
the problem.

In the example, nitrogen deficiency (1) leads to more
serious yield loss than does phosphorus deficiency
(2). Broad-leaf weed competition in beans (9) may be
moderately important, but researchers are not yet
certain of that conclusion (hence the "?"). Drought
stress (3) is responsible for sizeable yield losses in
maize (again, note that the number of farmers with
the problem was determined along with its
distribution). More information is required on the
frequency of drought: if it does not occur every year,
its ranking may drop, as is the case with
anthracnose in beans (6). Although it causes serious
yield loss, anthracnose occurs only one year in three,
and when both severity and frequency are
considered, the problem receives a medium rating
for seriousness.

Relative Importance of the Problems
When assessing the relative importance of problems
researchers should take account of all three criteria
(distribution, importance, and seriousness). Rankings
should be assigned for each criterion. In certain
cases some criteria might be given extra weight.
This procedure obviously provides only a rough
ranking, but it is an important start at setting
priorities for the on-farm experimental program. In
the example in Table 2, rankings are given using a
simple system of Xs and Os, in which the highest
priority is assigned to problems with the greatest
number of Xs, but other methods of reviewing the
rankings are also possible.

In the example, the problems of nitrogen deficiency
in tobacco (11) and waterlogging in beans (10) were
eliminated from further consideration. But note that a

decision to postpone or abandon research on a
particular problem does not depend only on its
ranking. It is also necessary to ask:

* How is a problem related to other problems? (See
Step 4.)
* Are solutions readily available? (See Step 6.)
* What resources are available for on-farm

As these factors are taken into account they can be
compared with the ranking given to problems in
order to decide the final composition of the
experimental program.


Step 2 ranks in rough order of importance the
problems that have been identified. Even problems
that have been only tentatively identified should be
considered here. If researchers have identified a large
number of problems they will eventually have to
eliminate some from immediate consideration. The
initial ranking carried out in this step is not precise,
but it will help researchers decide which problems
have a higher priority for the research program.

Problems should be ranked using well-defined
criteria. The criteria suggested here are: 1) the
distribution of the problem, including a definition of
which farmers in the research area are affected;
2) the importance of the crop enterprise to the
farming system; and 3) the loss of yield or income
for which the problem is responsible.

After the ranking in Step 2, problems that
researchers feel are of sufficient importance, and for
which sufficient evidence is available, are passed to
Step 3, where their causes are analyzed. Problems
that are potentially important but for which more
evidence is needed generally do not receive attention
in Step 3. Instead, researchers note the type of
experimental evidence (in List A) or information from
other diagnostic techniques (List B) that is required.

Fumner .'rJer..:rr
r.u',rcrj li, ,jenl ,lv
,r e a lu"lr cct llw rr.S

Step 3.
Identify Causes of the Problems

All of the problems identified in Step 1 that are
supported with sufficient evidence can be treated in
Step 3. Even problems that are only tentatively
identified may be analyzed in Step 3 if researchers
feel it is helpful. If some problems were assigned a
very low priority in Step 2, they need not be
considered in Step 3.

The object of Step 3 is to develop enough
information related to a particular problem to identify
appropriate solutions. The causes of a problem may
be management practices (or their absence-see
box, p. 43), natural or socioeconomic conditions, or
even other problems. Occasionally solutions can be
identified without knowing a great deal about the
causes of a problem, but in many cases lack of care
in defining causes limits the chances of identifying
reasonable solutions. Giving attention to the causes
of a problem is an important stimulus for identifying
imaginative solutions. If causes are not well
understood, it may be necessary to conduct
experiments or other investigations to clarify them.
The causes of problems examined in the on-farm
research program should be reviewed each year and
defined more precisely as more information becomes
available (see box, p. 41).

Diagramming Causes and Problems
The causes of problems may be fairly complex, and
it is often helpful to diagram the relationships
between causes and problems.

Diagrams for the entire example are shown in
Figure 2, pages 37-39, and discussed in the box on
pages 35-36. If there are several causes for a
problem, the diagrams may get rather large. Three
brief examples of diagramming problems and causes
follow on page 40.

Analysis of Causes in the Example

Figure 2 illustrates the analysis of the causes of the
problems that were identified in Step 1.

1. Nitrogen deficiency in maize is attributed to
several factors, including low levels of fertilizer
application; surface application of fertilizer, which
tends to be washed away with heavy rains; soil
erosion; and low levels of organic matter in the soil,
partly caused by farmers reserving crop residues
for forage.

2. Phosphorus deficiency in maize is judged to occur
because farmers do not apply phosphorus fertilizers.
Researchers also propose to test the hypothesis that
phosphorus is fixed in these soils when it is applied.
The nitrogen deficiency also contributes to the

3. Drought stress in maize at ear-filling time is
caused when the rains end early, especially in the
northern part of the research area. In addition, the
local variety, which farmers brought with them from
an area where rainfall is higher, matures quite late.
Moisture, retention is limited by the lack of organic
matter, which, along with soil erosion, is judged to
contribute to drought stress.

4. The high cost of weeding maize is caused by the
number of weedings (two to three) that farmers
perform and by the rapidly increasing cost of labor in
the area.

5. Nitrogen deficiency in beans occurs because
farmers do not apply nitrogen fertilizer to beans
(although they do apply phosphorus fertilizer). The
low level of organic matter in the soil and soil
erosion are additional causes of the problem.

6. Anthracnose attack on bean pods is caused by
local varieties' high susceptibility to disease, and
because farmers practice no control methods.

7.* The problem of root rots in beans is not yet
firmly established. Researchers speculate that if it is
confirmed as a problem it will be found mostly on
fields that are planted to beans every year. Those
are stony or sandy fields that farmers deem
inappropriate for other crops, and the lack of rotation
causes pathogens to build up.

8. Low plant population in beans is confirmed as a
problem but researchers are uncertain of the
causess. The problem may be due to root rots, if
they turn out to be present. Another possibility is
that the single tillage operation farmers perform
provides an inadequate seedbed, which leads to poor
stand establishment. A related hypothesis is that soil
crusting interferes with stand establishment. Finally,
farmers' lack of adequate seed storage methods may
lead to low quality seed that germinates poorly.

9.* Broad-leaf weed competition in beans is not yet
confirmed as a problem, but researchers have several
ideas regarding possible causes. They are 1) low
bean plant populations in most fields encourage
weed competition; 2) poor tillage practices do
likewise; and 3) most labor is devoted to
weeding maize.

36 Tentative problem

Figure 2.
Analysis of causes

of fertilizer

( Low doses of
fertilizer applied

Heavy rains
early in season

Soil erosionE

Low organic Farmers remove
matter in soil crop residues

Farmers use
crop residues
for forage

fixed by soil

No phosphorus
fertilizer applied

Exploratory experimentation on phosphorus placement is needed
because no economic response to broadcast fertilizer was noted
(List A).

SRains end

early in north

Soil erosion

Poor water ] Low organic
retention in soil matter in soil

SLocal variety matures too
late for the growing season)

Farmers brought the variety
with them when they settled
in the area 15 years ago /

(continued) 37

Figure 2.

Cost of labor
increasing rapidly

Soil erosion

Low organic
matter in soil

Local bean varieties
to anthracnose

Seed and soil
are not treated

Farmers do 2 or 3
weedings in maize

No nitrogen
fertilizer applied

Extension does
not recommend
nitrogen for beans

No chemical
disease control

Many bean fields
not rotated

Certain sandy or
stony fields are
used only for beans

*Tentative problem

Poor seed storage

Poor seed quality

Soil forms crust

Heavy rains
early in season

Poor seedbed

Only 1
superficial tillage

Further investigation is needed on possible causes of !ow plant
population in beans (List B). Seed quality-Germination tests on
farmers' seed. Soil crusts and seedbed- Observe emergence rates
with farmers' soil conditions and land preparation practices.
Root rots-Observations and laboratory tests.

SOnly 1 hand
weeding in beans

Farmers do 2 or 3
weedings in maize


*Tentative problem

*Causes of causes-Often it is necessary to give
more information about a particular cause, and
doing so may create chains of causes. One of the
causes of nitrogen deficiency in maize is low
organic matter, which in turn is partially due to
farmers' practice of removing crop residues and
using them for fodder. The diagram for the causal
chain looks like this:

Farmers Farmers Low
use crop remove organic
residues for crop matter
forage residues

* Multiple causes-More than one cause may be
identified for a particular problem. If several
factors contribute to a problem, all of them should
be presented. The diagram may be quite
complicated, as is the case with nitrogen
deficiency in maize (Figure 2).

If two causes act together, the arrows can be
joined, as shown in the diagram where the surface
application of fertilizer is particularly inappropriate
because heavy rains wash the fertilizer away.
When a cause is uncertain, the relationship can be
indicated with a question mark (?). If it is later
shown that this cause is unimportant, it should be
removed from the diagram.

* Problems as causes-Sometimes two problems are
related to each other, in which case it is important
to specify how one problem contributes to the
other. Nitrogen deficiency is seen as a contributing
cause of phosphorus deficiency in maize (2)
(Figure 2).

Definitions of Problems and Causes

Sometimes the difference between problems and
causes is unclear. In the example, erosion is
presented as a cause of nitrogen deficiency and
drought stress. It may be argued, however, that
erosion itself is a problem, representing a serious
inefficiency in resource use. There would be no
difficulty in treating erosion as a separate problem
that is a contributing cause of other problems.

The definitions of problems and causes occasionally
change as on-farm research progresses. For the
problem of low plant population in beans (8), several
causes are proposed. They include poor seedbed
preparation, poor seed quality, soil crusting, and root
rots (itself a problem). If it is found that poor
seedbed is the principal cause, the analysis can be

SOnly one Poor
perficial tillage seedbed

If it is found that root rots are the principal cause of
low plant population, the description of the problem
will change. Low plant population can simply be
considered a symptom of the root rot problem and
there will be no separate diagram for low plant

This sort of adjustment is normal as problems and
causes become better defined during the course of

Limitations in Identifying Causes
The number of causes-The number of causes or
possible causes listed for a problem should be
limited. They should not include wild guesses, nor
should they extend the causal chain to extremes. If
there is evidence that a foliar insect is damaging a
crop, it is probably not worthwhile to relate the
physiology of the insect to the damage that it does.
But researchers do want to ask if the insect's
occurrence is associated with a particular rotation,
time of planting, or other possible contributing cause
that would help them consider alternative solutions
to the problem. Natural conditions such as wind or
rain may be listed as causes if they help in
considering possible solutions. A useful rule is that
only enough information regarding causality should
be developed to help researchers think of practical
solutions to the problem.

When researchers consider the type of diagnostic
work required to identify causes, they may find it
helpful to list all possible causes and then work
toward eliminating the less plausible ones. Diagrams
with many hypothesized causes are not very useful
because excessive detail and extraneous causes lead
to unfocused analyses. On the other hand, including
too little information limits the possibility of finding
imaginative solutions.

Problems that do not require causes to be listed-It
is sometimes the case that no causes need to be
provided for a particular problem, as in the example
of the foliar insect problem. If the climate and other
conditions are appropriate for the insect and if no
management factors or other circumstances which
might exacerbate insect attack (and lead researchers
to think about possible solutions) are identified,
then there is no need to diagram any causes for
the problem:

The Absence of a Practice as a Cause

Farmers' practices are commonly identified as causes
of problems. Late fertilizer application may be one
cause of a nutrient deficiency, for instance. But
what if farmers don't use any fertilizer, as is the
case with nitrogen deficiency in beans (5)? Can this
"non-practice" be used as a cause? Sometimes it is
useful to explain a particular problem by noting that
farmers do not do something and, where possible, to
explain why:

Extension does No nitrogen
not recommend fertilizer
nitrogen fertilizer applied
for beans applied

But it is important to realize that fertilizer is not the
only means to overcome the nutrient deficiency. The
danger of including non-practices as causes is the
tendency to limit the search for solutions. The fact
that farmers do not use fertilizer helps explain the
presence of the problem, but researchers should
inquire if there are other causes that suggest
alternative solutions.

If researchers are considering why farmers do not do
something, care must be taken in listing the causes.
For instance, if it appears that a crop can be planted
at a particular time of year, but farmers leave the
land idle, possible causes should include only those
reasons expressed by farmers (e.g., lack of labor) or
obvious to researchers (e.g., adverse climatic
conditions). Speculation as to what might happen
should a crop be planted-for example, nutrient
deficiencies might develop-should take place when
evaluating possible solutions (Step 6, criterion 1) and
not when analyzing causality.

Further evidence required-If a problem is not
considered sufficiently important it should not be
analyzed in Step 3 (so problems 10 and 11 do not
appear in Figure 2). Likewise, if a problem was not
well defined in Step 1, it need not be considered in
Step 3. Although problems 7 and 9 are not yet
confirmed, they appear in Figure 2 because
researchers wish to consider the possible causes of
the problems. But if a problem is not well defined,
proposing research to explore possible causes is
rarely worthwhile until the problem is confirmed.

Sometimes researchers will decide that they need
more evidence regarding the causes of a problem
before proceeding to consider possible solutions. In
Figure 2, two such cases are illustrated. Problem 2
(phosphorus deficiency in maize) requires more
information regarding the fixation of phosphorus, and
an exploratory experiment on fertilizer placement is
proposed. The experimental factor should be
included in List A. Problem 8 (low plant population in
beans) may be due to one or more causes. The
proposed investigation includes germination tests on
the seed used by farmers and observations on
emergence rates in farmers' fields. These diagnostic
techniques should be noted in List B.


The third step in the planning process is to identify
the causes of problems for which there is sufficient
evidence (from Step 1) or for which it would be
useful to analyze causes. The cause may be farmers'
natural or socioeconomic circumstances or cultural
practices. This step should only be done for
problems that researchers believe are important
enough to deserve attention (from Step 2).

Because the causality of problems is sometimes
quite complex, it is helpful to diagram causes and
problems by using an arrow to lead from causes to
problems. In some cases a chain of causes may lead
to a particular problem, or in other cases several
causes may contribute to a problem.

The evidence for the causes of each problem may
come from previous experiments, surveys, or other
diagnostic techniques. Researchers should decide if
they know enough about the causes of the problem
to go on to consider possible solutions, or if they
need more evidence to identify or confirm the

When the causes have been listed, they are passed
to Step 4, where interrelations among problems and
causes are considered.

F url hF F E .aE nC

3or ba-. r'.
Identify causes

SUat Ahb .
Factors for Otheir.
lexperimentationl chagnost

Step 4.
Analyze Interrelations
among Problems and Causes

To help select priority problems and consider which
factors might be examined in the same experiment,
it is useful to review the interrelations of problems
and causes identified in the previous steps. The
review should include all well-defined problems from
Step 1 that are of high enough priority, along with
their known and possible causes.

One way to make interrelations among problems and
causes evident is to try to combine the individual
diagrams of problems and causes from Step 3 into a
single summary diagram. Each problem and cause
appears only once in the new diagram. If there are a
large number of problems and causes, arranging the
summary diagram may take several drafts. There is
no single "correct" diagram; it is simply an aid to
visualizing the interrelationships. The example is
presented in Figure 3, pages 48-49.

When there are many problems and associated
causes, one summary diagram may be too large and
complicated to be useful. In this case smaller, partial
summaries examining problems that exhibit strong
interrelations should be considered. If different
problems are associated with different groups of
farmers, separate summaries may be called for. If
the research program is studying more than one crop
enterprise, separate summaries for each could be
developed if there are no interactions between the
enterprises. In Figure 3 the summaries for the two
crops are presented together because of two
interactions. Labor for weeding maize affects
weeding in beans, and the two crops also share the
problem of nitrogen deficiency. As research
continues, other interactions involving the
management practices used by farmers in growing
the two crops might be uncovered, especially when
the crops are grown in rotation.

Analyzing interrelations is useful not only for helping
decide which problems and causes should receive
more attention, but also as a reference later in the
planning process when researchers consider the
design of experiments. Recall that problems that are
still not well defined need not appear in this analysis.

Although root rots and broad-leaf weed competition
in beans are not yet confirmed as problems, they
appear because of possible interactions with other

In carrying out the analysis, it is helpful to pay
attention to four kinds of associations among
problems and causes.

* A particular cause is involved in more than one
problem. Such causes may deserve extra attention
when considering possible solutions because they
may offer possibilities for resolving several
problems at once. One example of that situation is
low soil organic matter as a cause of three
problems: nitrogen deficiency in maize and beans
and drought stress in maize. Improving soil organic
matter may help solve all these problems. Another
example is that reducing the amount of labor for
weeding maize might not only lower the cost of
weeding maize but also make more labor available
to improve broad-leaf weed control in beans.

* Two problems are interrelated. In this case it is
necessary to ask if one problem must be resolved
before work begins on the other. Farmers are
likely to change their practices in steps, and the
sequence of those steps must be considered in
deciding which problems and causes deserve
attention first. A solution to the nitrogen
deficiency problem is required before, or
concurrently with, work on phosphorus deficiency
in maize. Although a number of the relationships
are not yet established, Figure 3 provides material
for further speculation. For instance, if it is
confirmed that broad-leaf weed competition is a
problem in beans and that one of its principal
causes is low plant population, the possibility
exists that weed competition might be at least
partially reduced by something as seemingly
unconnected as improvements in seed storage.

* A problem has several contributing causes. In
such cases, the causes may best be examined in
the same experiment. In Figure 3, two causes that
contribute to the root rot problem are the lack of
crop rotation and the lack of seed or soil
treatment. If both causes suggest possible
solutions to the root rot problem, those solutions
should be tested in the same experiment.

Figure 3.
Problems and causes in maize and beans

Surface Heal, r3.n
applcaton ear n season
of fertilizer

1 "


LOaw doses
of fertilizer

LawN organic
marler in sail

1fIxed br/sa. JveIt


croa residues

Farmers use
crop residues
foi forage

Poor water
retentron~ rn

Rains end early I
in the north

Local aarsert
matures too late
for he growing

Farmers brought
the variely witn
them iwhen they
sertled the area
15 years ago

Cost of labor

Farmers do 2
or 3 hand
in maize


Soil forms

Poor seed

Poor seed

( No nitrogen
fertilizer applied

Seed and soil
are not treated


OnIe 1
suoerlic aIJ

bean fields
not rotated

Certain sandy
or stony fields
are used only
for beans

Extension does not
recommend nitrogen
for beans

Local bean
susceptible to

Only 1
hand weeding
in beans

No chemical
control used

* Tentative problem

* A problem has no causes in common with other
problems. Research on such a problem may
proceed independently. Anthracnose appears to be
a case in point, so that experiments to examine
the problem need not include any other
experimental factor.11


The fourth step is an attempt to examine the
interrelations among the problems and causes that
have been identified. Very often problems are related
to each other, either directly or through shared
causes. This step allows researchers to see those
relationships and to think about their implications.

The best way to examine interrelations is to try to
combine the causal diagrams for each problem into
one complete diagram. If the research is studying
different crops or different groups of farmers that
have nothing in common, separate diagrams for
those crops or farmers may be drawn.

An examination of the overall relationship is helpful
for thinking about research priorities. A cause that is
related to more than one problem may deserve extra
attention where solutions are proposed. If one
problem contributes to another, the first may need to
be addressed before, or concurrently with, the
second. If a problem has several contributing causes,
those may be addressed in the same experiment.
Finally, if a problem has no causes in common with
other problems, it may be addressed separately in
the experimental program.

11 Unfortunately, there is an interaction in this case. It is
probable that increasing plant population will increase the
anthracnose problem, as the effects of anthracnose tend to
become more severe with higher plant population. So
experimentation on these two problems should be
considered together. See p. 57.

After these interrelations are examined, the problems
and causes that researchers consider to be
sufficiently important are passed to Step 5, where
solutions are proposed. Problems whose causes are
not well defined are generally not considered in Step
5. Instead, researchers note the type of experimental
evidence (in List A) or information from other
diagnostic techniques (in List B) that is required.

Identify causes

Further evidence
required to identify
or evaluate problems

Analyze interrelations
among problems
and causes
Further evidence
required to determine
causes of problems

List A
Factors for

List B

Step 5.
Identify Possible Solutions to the Problems

In Step 5 possible solutions to the problems are
identified. This step can only be taken when
researchers have enough confidence in the evidence
available for a problem and its causess. Thus only a
few problems are presented in Table 3: nitrogen
deficiency in maize (1), drought stress in maize (3),
the high cost of weeding maize (4), nitrogen
deficiency in beans (5), and anthracnose in beans
(6). All are problems with sufficient evidence of their
causes. Recall that other problems have not been
abandoned, but are treated in List A for exploratory
experimentation or in List B for additional diagnosis.

Researchers should try to consider as broad a range
of solutions as possible in this step (in Step 6 the list
of solutions will be narrowed). The proposed
solutions may be inputs, crop varieties, cropping
patterns, or cultural practices. They should be
specified as clearly as possible (e.g., type of
herbicide), but exact dosage or levels will be
determined when designing the experiment.

Researchers should devote considerable time to
brainstorming when they develop ideas about
possible solutions. It is often useful to list
technologies available for the area. Participants in
the planning session should review the work of the
local research service in breeding and crop
management, and consider other innovations
reported in the literature. In Table 3, solutions 6a
(new bean varieties) and 6d (lines with greater
market acceptability) were suggested by local bean
breeders based on progress in their varietal
development program. Solution 6b (products for
foliar disease control) came from some work that
pathologists did on the experiment station.

In considering solutions to problems, the place to
start is their causes. Causes help suggest ways of
attacking problems. In the case of drought stress in
maize (3), not only is an earlier maturing variety
considered a potential solution, but analysis of the
causes suggests a possible solution in the form of an
intercrop to control erosion. Examining the causes of
a problem may help rule out some possible solutions
as well. When it is considered that one of the causes

Table 3
Identify possible solutions to problems (Step 5)

Nitrogen deficiency
in maize

Drought stress in maize
at ear filling

High cost of weeding
maize by hand

Nitrogen deficiency
in beans

Anthracnose attack on
bean pods

a) Apply 80 kg N/ha, half at planting and half at 30 days,
in hole at side of maize (this solution has been tested with
success in previous experiments).
b) Incorporate maize residues to build up organic matter
in the soil.
c) Purchase chicken manure and apply to fields.
d) Plant leucaena strips to control erosion and supply nitrogen
through leaf litter and periodic pruning for mulch.

a) Plant early maturing maize variety A, which has been
tested for two years and is ready for recommendation.
b) Plant leucaena strips for erosion control.

a) Apply pre-emergence herbicide C and postpone first
weeding to 40 days.
b) Apply pre-emergence herbicide D and postpone first
weeding to 40 days.
c) Apply pre-emergence herbicide E and postpone first
weeding to 40 days.

a) Apply nitrogen fertilizer.
b) Purchase chicken manure and apply to fields.
c) Inoculate seed with rhizobium.
d) Plant leucaena strips to control erosion and supply nitrogen
through leaf litter and periodic pruning for mulch.
e) Incorporate maize residues to build up organic matter
in the soil.

a) Plant tolerant bean varieties J, K, and L.
b) Use mixture of fungicides M and N.
c) Use mixture of fungicides P and Q.
d) Plant 10 anthracnose-tolerant bean lines (more marketable
seed type).

The following problems are not considered for possible solutions:

Phosphorus deficiency in maize (2). The cause of the deficiency is not clear. Exploratory
experimentation will see if phosphorus is being fixed by the soil. (See List A.)

Root rots of beans during crop establishment (7). The importance of the problem is not yet
certain. Field observations and laboratory tests will be carried out. (See List B.)

Low plant population in beans (8). The cause of the problem is not clear. Field observations
and germination tests will be carried out. (See List B.)

Broad-leaf weed competition in beans (9). The importance of the problem is not yet certain.
Exploratory experimentation will measure the yield loss from weed competition. (See List A.)

of nitrogen deficiency in maize (1) is the method of
applying fertilizer, it becomes obvious that simply
increasing the dosage is not a possible solution.

In some cases a solution to a problem will already
have been tested in the experimental program in
previous cycle(s), and sufficient evidence of its
success may mean that it is ready for demonstration
to farmers. This is the case for solution 3a, a new
maize variety that has performed very well in the
past two years' experiments. Solution la (fertilizer
application) has also been tested for two years and
is ready for a final verification.


The fifth step in planning on-farm experiments is to
list solutions for those problems for which
researchers have sufficient evidence and whose
causes are understood well enough to suggest
possible solutions. Possible solutions for each
problem should therefore take into account what
researchers know about the causes. Researchers
should note any solutions that they think might be
feasible based on research conducted by their
institutions or reported in the literature, or based on
their experience. Each of the proposed solutions will
be evaluated in Step 6.

Further evidence
required to identify
3 or evaluate problems
IdenTivy causes

I------* ------Iideti

Analyze inierrelations
among problems
and causes
Further evidence
required to determine
causes of problems
Identify solutions
Uti A List B
Factors for ; Other
experimentation diagnostic

These criteria are presented in rough order of
importance. Criteria 1 and 2 are the most crucial. If
researchers have evidence that the proposed solution
will not function or will not be profitable, then it
should be eliminated from consideration. Criteria 3,
4, and 5 are also quite important. If solutions appear
to be incompatible with the farming system, increase
risks for farmers, or require much institutional
support, researchers should consider whether it is
likely that the solutions will be adopted. Unless a
solution offers great advantages, it will be better to
look for alternatives. Criteria 6 and 7 are rarely
sufficient in themselves for eliminating a proposed
solution, but in combination with other criteria they
may suggest that another solution to the same
problem deserves higher priority.

These seven criteria are presented only as
suggestions; researchers may prefer to emphasize
some others. But three important factors should be
addressed when developing any criteria for
evaluating possible solutions:

* The potential benefits of the solution to farmers
(an issue addressed here by criteria 2 and 4);
* The ease with which farmers can adopt the
solution (criteria 3, 5, and 6); and
* The ease of investigation (criteria 1 and 7).

Researchers should rank each potential solution using
the set of criteria they have developed. Next, they
should review the ranking and judge the value of
conducting future research on the potential solution.
Potential solutions should be reviewed each cycle.
The experimental results of previous cycles will
suggest whether certain solutions should be
promoted, retained for further experimentation, or

An example of the evaluation of possible solutions is
presented in Table 4.

1) Probability That the Technology Will Function
If a proposed solution has been included in on-farm
experiments in previous years, researchers can judge
its performance. But if it has not been tested before,
researchers must ask themselves how certain they

Step 6.
Evaluate Possible Solutions

Because experimentation is the most costly phase of
OFR, researchers must make sure that the possible
solutions included in the experimental program have
a high chance of success. In Step 5, many possible
solutions to a given problem were considered. In
Step 6, the list of solutions will be narrowed by
evaluating each solution according to seven criteria:

1. Probability that the technology will function-
Researchers must consider whether it is likely
that the technology will function under the
agroecological conditions and management
practices of target farmers.

2. Profitability-Farmers will not be interested in a
new technology unless it is profitable.

3. Compatibility with the farming system-Solutions
to farmers' problems should be compatible with
the other elements of the farming system-the
socioeconomic and natural circumstances,
management practices, and other crops, animals,
and off-farm employment managed by farmers.

4. Contribution to reducing risk-Farmers will be
most interested in solutions that help reduce risk
in their farming operations.

5. Need for institutional support-Researchers
should assess whether the proposed solution will
require special support from extension services,
the provision of new inputs, or a change in
credit programs.

6. Ease of testing by farmers-There is a better
chance of farmers using a technology if they can
test it for themselves without a high initial
investment of cash or labor.

7. Ease of carrying out the experimental program-
All other things being equal, solutions that can
be tested at low cost are preferable to those
that require very expensive experimentation.


are that it will work under farmers' circumstances
and practices. Technologies that work in other areas
or on the experiment station are not necessarily well
adapted to the local situation.

Sometimes a technology may have unintended
consequences, as when a new method of weed
control increases erosion on hillside fields.
Researchers want to take such factors into account.
In other cases, they may conclude that a proposed
solution will only be successful if additional changes
are made (for example, the use of a herbicide may
require changes in timing or type of cultivation). In
such instances the additional factors should be listed
as part of the proposed solution and included in the
rest of the evaluation.

Occasionally the resolution of one problem may
actually worsen another, as with plant population
and anthracnose in beans: if plant population is
increased, the severity of anthracnose will be
greater. The problem of low plant population has not
been passed to Step 5 because its causes are still
being investigated. But researchers should be certain
that they have some way to address the anthracnose
problem before they resolve the problem of low plant

Each possible solution should be evaluated based on
the probability that it will function in the local
situation. A number of possible solutions in Table 4
get a high rating, either because of results in earlier
on-farm experiments or because researchers are
familiar with their performance in similar situations.
Lack of experience with solutions 1b, 1c, ld, 5c12,
and 6d gives them medium rankings. Fungicide Q
(6c) has been found to be toxic for beans if the
proper dosage is exceeded. In this case, a low rating
is sufficient to eliminate fungicide Q from the
experimental program.

12 There is some concern that solution 5c, rhizobium, may only
function in the presence of added potassium, so it receives a
medium ranking and researchers suggest that potassium be
added to List A.

Table 4
Evaluate possible solutions (Step 6)

Possible solution

la 80 kg N/ha for maize,
half at planting and
half at 30 days, in

lb Maize residues

Ic Chicken manure

ld Leucaena

3a Maize variety A







Profitability system







(farmers use
residues for




3b Leucaena (See id)

4a Pre-emergence
herbicide C

4b Pre-emergence
herbicide D

4c Pre-emergence
herbicide E

(but requires

(but requires

(but requires












(very toxic
to humans)





x x

x x

x x

Medium High Verify on large plots
with farmer

Medium Low Postpone until other
sources of fodder
are developed

Medium Medium Eliminate-not

Low Low Begin research to
develop technology

High High Demonstrations

Medium High Experiments-
(need to possible solution

Medium High Eliminate
(need to

Medium High Experiments-
(need to possible solution




Table 4

I Nitrogen fertilizer for


Medium High


U Chicken manure (See lc)

I Rhizobium

(may need



SLeucaena (See d)

-n Maize residues (See
1 b)

U Bean varieties High Medium Medium High
J, K, and L (only those
farmers who do
not market

I Fungicides M, N High Medium High High

Fungicides P, Q Low Q Medium High High
is very toxic
to beans if
dosage is

U Ten new bean lines





t Anthracnose severity is likely to be greater with high plant populations (see p. 76).

x Medium

High Experiments-
possible solution

Medium High Experiments
(include K)

High High Experiments-
possible solution

Medium Medium Experiments-
possible solution

Medium Medium Eliminate


High Experiments-


x x

/I I/

0 IN

2) Profitability
An estimate of the profitability of each proposed
solution should be provided. If a solution has been
tested in a previous cycle's experiments, the
economic analysis of the results should give a good
idea of its economic viability. If the proposed
solution has never been tested on farmers' fields, an
estimate of its profitability should still be attempted.
Making the estimate involves assessing all of the
changes in costs that the farmers must incur in using
the new technology, and comparing those costs with
an estimate of the yield difference that farmers can
expect when using the technology under their
conditions. Solutions that researchers believe have
little chance of being profitable at present or in the
future should not be tested on farmers' fields.

The profitability of each possible solution is
presented in Table 4. Solutions la and 3a were
included in on-farm experiments in previous cycles,
and an economic analysis of the results provided a
good estimate of profitability. For solutions that had
not yet been tested, researchers estimated
profitability by comparing an estimate of likely yield
increases with the increase in costs that the new
technology represents to farmers (see box,
pp. 66-67). The analysis of profitability for the new
bean varieties (6a) included the fact that although
the new varieties yield more, their market value is
not as high as that of the traditional varieties. In the
case of herbicides for maize (4a, b, and c), yield
changes are not expected, but lower production
costs are anticipated. An economic analysis of the
application of chicken manure (1c) showed that its
profitability would be quite low and it was eliminated
from further consideration. Finally, three possible
solutions (1b, 1d, and 5c) presented so many
technical uncertainties that researchers could not
estimate profitability.

Estimates of profitability can sometimes be quite
complicated and will require special assistance from
economists. This is especially true for technologies
that offer benefits over the long term, such as
technologies to improve soil fertility or structure, or
methods of controlling erosion. Researchers should
consider the trade-offs involved in choosing
experimentation aimed at providing solutions whose
benefits will only be apparent after a number of

years over choosing to explore solutions that provide
more immediate returns. In many instances a number
of short-term opportunities will take precedence,
whereas in other instances important long-term
considerations, such as sustainability and/or a lack of
available technologies, will dictate a different
direction for the research program.

The proposal to experiment with planting leucaena
strips is a good example. Although the effects of
such a solution will probably not be evident for
several years, it may make important contributions to
improving soil fertility and moisture retention.
Researchers believe that solution's potential impact
is great enough to justify initiating some
experimentation with leucaena.

3) Compatibility with the Farming System
In a diagnosis, researchers are interested in
understanding the reasons for farmers' practices to
better define the scope for proposing new
technologies. As researchers form some ideas of
possible solutions, each should be analyzed for
compatibility with the farming system (see box,
p. 69). During a survey, for example, farmers might
be asked what they think of a particular technology,
why they don't already use it, or what they think
might happen if they did use it.

Thus an important aspect of planning involves a
comparison of proposed solutions with what is
known about farmers' circumstances. For example, a
new variety is proposed for subsistence farmers. Is it
acceptable for the family's food preparations? A
change in planting methods is proposed that requires
more labor. Is extra labor available at this time of the
year? A new herbicide is proposed. Can it be used in
the current rotation pattern? Researchers must
review what they know of the management
requirements of other crops and animals in the
farming system, the land and labor resources
available to the farm family, and the family's goals
and preferences, to decide if a proposed solution
is compatible.

There are two examples in Table 4 of solutions that
are incompatible with farmers' circumstances. The
first-incorporating maize residues to increase soil
organic matter (1b)-would interfere with their use

as animal fodder. Unless researchers can propose an
alternative source of fodder, or unless they feel that
the extra yield from fields with more organic matter
would compensate for the economic value of the
fodder that is lost, it is not worthwhile to pursue
that solution.

The second solution that is incompatible with
farmers' circumstances is the use of herbicide D
(4b). Researchers know that it is very toxic and
because farmers in the research area do not have
much experience managing chemicals it is felt best
to eliminate this product from consideration.

A proposed solution may be compatible with the
circumstances of some farmers but not others, a
situation which should be noted. In such cases the
particular group of farmers for whom the solution is
intended should be described. For example, bean
varieties J, K, and L (solution 6a) are acceptable for
home consumption but receive a heavy discount in
the market. Thus they are only appropriate for small-
scale farmers who do not market their beans.

A low rating on compatibility with the farming
system does not necessarily eliminate a possible
solution from consideration. The basis for on-farm
research, after all, is the conviction that farming
systems can be improved. Rather than looking at the
system as absolutely fixed, researchers should use
their knowledge of its characteristics to ask, "Do we
understand the trade-offs involved in choosing this
solution?" In the example of incorporating maize
residues to increase soil organic matter (lb), the
trade-off can be described as weighing the rationale
for the current practice (feed for animals, ease of
land preparation) against possible gains from the
alternative practice (improved soil fertility and
structure, higher yields). In this case, the judgment is
that the gains from the change will not compensate
the losses to the current system, but if researchers
find an alternative source of fodder that decision
may change.

The new bean varieties (6a) are only a compatible
solution for farmers who do not market beans, and
research on the new varieties will depend on how
many farmers do not market beans. If the number is
very small, the solution may be abandoned.

However, researchers already know of anthracnose-
tolerant bean lines that are acceptable in local
markets. They know little about their adaptation to
farm conditions and so decide to test 10 lines to
increase the chances of success (solution 6d).

4) Contribution to Reducing Risk
Risk is an important determinant of farmers'
practices: farmers may stagger planting dates to limit
losses from drought, or plant several different
varieties to reduce losses from disease. Such
practices may indicate important problems that merit
researchers' attention. Risk is also a concern when
farmers consider adopting new practices.

Researchers should ask whether it is likely that
farmers will face possible losses in some years if
they adopt a proposed solution or if the solution is
likely to increase the stability of farmers' incomes.
If experimental evidence from previous cycles is
available, the variability in results will help
provide some indication of the risk involved in a
proposed solution.

In Table 4, possible solutions that contribute to
reducing risks caused by drought or disease (3a,
6a, b, c, d) get high rankings. An improvement in
soil organic matter (1b, c, d) would certainly
contribute to stability as well. Pre-emergence
herbicides (4a, b, c) present a bit of risk because
they require farmers to make most of their
investment in weed control at the beginning of the
season; as it is now, farmers adjust their hand
weeding investment according to the rains and the
growth of the crop. Finally, an analysis of previous
experiments with fertilizer (la, 5a) and a knowledge
of drought risk in the study area lead researchers to
exercise caution with respect to this solution,
especially if it is used in the northern part of
the area.


if a proposed solution to a problem requires higher
costs than the farmers' current practice, then. an
assessment of the returns to those additional costs
must be made. This evaluation is an important part
of analyzing on-farm experiments. One way to do it
is to draw up a partial budget and then carry out a
marginal analysis for the costs that vary and net
benefits of each treatment.13 The partial budget

* The gross benefits for each treatment (the yield
multiplied by the field price of the product);
* The total costs that vary for each treatment; and
* The net benefits (gtoss benefits. minus total costs
that vary).

The marginal analysis examines changes in costs
that vary and net benefits between treatments. In
the simple case of an experiment with two
treatments (the farmers' practice and an alternative),
the change in net benefits is divided by the change
in costs that vary to give a marginal rate of return.
That rate of return must be above the minimum rate
of return acceptable to farmers (typically
50%-100%) before the alternative can be

These same ideas can be used to evaluate proposed
solutions before they are included in experiments.
The additional costs due to the proposed solution
and the field price of the product can be used to
calculate a minimum yield increase acceptable to
farmers. Researchers can then judge whether the
solution can reasonably be expected to produce such
a yield increase.

Proposed solution 1c, chicken manure, can serve as
an example. First, calculate all of the additional
costs. In this case, researchers propose that farmers
apply the chicken manure in addition to their current

13 A method of economic analysis for on-farm research Is
provided in CIMMYT (1988).

fertilizer application. The additional costs are thus all.
of the costs (expressed as $/ha) associated with the
chicken manure:

Cost of 8 tons of chicken manure $ 96
Cost of transporting manure to farm 12
Cost of applying and incorporating manure 20
Total increase in costs
for proposed solution $128

Second, remember that farmers not only want to
recover the extra costs they have incurred but also
expect a return on that investment. If the minimum
rate of return is 50% (i.e., for every $1 invested,
farmers want to recover the $1 and an additional
$0.50), then the minimum return expected by
farmers is 1.5 x $128, or $192.

Third, it is necessary to estimate the field price of
the maize-what an extra kilogram of maize is worth
to farmers. The field price is the price that farmers
can receive for selling the maize, minus all of the
costs proportional to yield that are associated with
harvesting and selling the grain. In this case, the
field price of maize was calculated to be $0.12/kg.

Finally, the minimum return is divided by the field
price to determine the minimum yield increase
expected by farmers:

$ 192
$ = 1,600 kg

In this case, researchers judged that the application
of this quantity of chicken manure is very unlikely to
give a yield increase as high as 1,600 kg of maize
per hectare. For that reason, they decided to
eliminate this proposed solution from further

Although these calculations will not always give
clear answers about the feasibility of proposed
solutions, it is very important that researchers have a:
good idea of the changes in costs implied by each of
the solutions that they are proposing before including
them in on-farm experiments.

Low ratings on the contribution to reducing risk
require careful consideration. Researchers will have
to balance the gains of a new technology against the
risks entailed, and ask if farmers are able to bear
those risks. In the case of nitrogen fertilization (la),
a low rating must be weighed against the high
(average) profitability observed in experiments.
Researchers will have to analyze carefully the
financial risks for farmers before making the nitrogen
recommendation. Part of the risk of the technology
is related to drought, and as researchers believe that
they have identified some solutions to the drought
problem, there is hope that nitrogen fertilization will
not involve very high risks for farmers.

5) Need for Institutional Support
Extension should be a part of on-farm research
activities and certain experiments may be used by
the extension agency to demonstrate new
technologies to farmers. Some proposed solutions
will be accessible to farmers without any special
institutional support, whereas others will require
extra support from extension in training farmers. In
the example, cases include: splitting the nitrogen
application in maize (la); incorporation of maize
residues (1b); planting leucaena strips (1d); the use
of pre-emergence herbicides to control weeds in
maize (4a, b, c); and rhizobium inoculation in beans

Some proposed solutions may require inputs that are
presently unavailable in the area. Researchers must
decide if there are other options, or if it is worth
communicating with those who are in charge of
supplying inputs to assure that a particular input will
be available to farmers. Recommending an input that
farmers cannot obtain is a waste of time. Herbicide E
is not currently available in the area, for instance. If
initial experiments find it to be promising, then
researchers will have to enquire about the possibility
that it can be made available. If the new maize
variety (3a) is to be taken further after successful
trial results, researchers must make sure that seed
production and distribution are being considered. If
the bean varieties (6a) or lines (6d) prove successful,
seed supply will be important as well. Researchers
must also consider the availability of planting
materials for leucaena (1d) and of rhizobium
inoculants (5c).

Compatibility with the Farming System
The concept of farming systems has received
increasing attention in recent years. It is based on
the recognition that many farmers manage a rather
complex set of crops, animals, and off-farm
enterprises, and that their choices about what to
grow and how to grow it are conditioned by both
natural and socioeconomic circumstances and by the
constraints imposed by other elements of the
system. Although researchers may focus their
attention on one or a few crops within the system,
they must be aware of how practices applied to
those crops are influenced by farmers' circumstances
and the management requirements of other elements
of the system. When possible solutions are
evaluated, their compatibility with the rest of the
farming system must be assessed. Two examples

In one study area, weed competition in beans was
seen as a serious problem. A proposed solution was
that farmers carry out one extra hand weeding, but
an analysis of the farming system showed that labor
was very scarce during the time that the extra
weeding would have to be performed. At that time
most farm families were busy picking coffee on
neighboring plantations. Coffee was a very important
source of cash income to the farmers, and it was
unlikely that the returns from an extra hand weeding
in beans could compete. Thus alternative solutions to
the weed problem had to be considered.

In another study area, a problem of moisture
deficiency during the growing season for wheat led
researchers to propose very early tillage to conserve
moisture. But animals were an important part of the
farming system and grazed on the fields until just
before planting time. The early tillage solution was
therefore not very compatible with the farming
system, and the advantage of early tillage had to be
balanced against the constraints imposed by the
farmers' management of their animals.

Occasionally a proposed solution may require that
farmers have access to credit, and again researchers
must determine if the solution is feasible and if it is
worth talking to officials in charge of credit policy. In
the study area some credit is available to farmers to
obtain fertilizer for maize and beans. If more fertilizer
is to be recommended, it might be helpful if more
credit were available.

When a solution requires institutional support,
its feasibility should be investigated before
experimentation proceeds very far. If a certain input
cannot be obtained, it should be discarded as a
possible solution. None of the proposed solutions in
the example is ruled out because of this factor, but
several require that researchers investigate the
availability of inputs, credit, or extension programs.

6) Ease of Testing by Farmers
Farmers are more likely to be interested in solutions
that they can try out a little at a time, especially if a
considerable investment is involved. For example,
farmers will be less interested in trying a solution
that requires the purchase of a new implement than
in trying an input that they can buy in small
quantities. However, this consideration should not be
used to rule out possible solutions that require a
large investment. With respect to machinery, for
example, a few individuals may invest in a new
implement and then develop a rental market.

Table 4 lists several possible solutions that farmers
can easily test a bit at a time, including new
varieties (3a and 6a, d). Other possible solutions are
slightly more difficult to test gradually. Planting
leucaena strips requires farmers to make a
considerable commitment initially, and it gets a
"low" rating. This is not sufficient to eliminate it
from consideration, but researchers must be aware
that it will be less easy for farmers to adopt this
solution than some alternatives.

Farmers also prefer to change their practices in
steps. Whenever possible, technologies should be
tested so that a series of changes rather than an
"all-or-nothing" package can be offered to farmers.
Nevertheless, sometimes there is no alternative but
to propose a combination of changes, as when a
new tillage method also requires changes in weeding

practices, or a new variety requires a different
planting density. But the more complex the
recommendation, the more difficult it will be for
farmers to adopt quickly. Three proposed solutions in
Table 4 imply some complexity. The change in
fertilizer practices (la) includes both a change in
dosage and method of application. The use of pre-
emergence herbicides (4a, b, c) requires farmers to
postpone their traditional weeding from 30 days to
40 days to avoid breaking the herbicide film. Neither
solution is so complex as to be unacceptable, but
each is an example of the factors that have to be
taken into account when considering an innovation's
acceptability to farmers. As for rhizobium inoculation
(5c), researchers are concerned that it may only
function in combination with the application of
potassium. If this were to be the case, it might have
to be eliminated.

7) Ease of Carrying Out the Experimental Program
Some proposed solutions are more costly than others
to investigate and therefore may receive lower
priority in the experimental program. Long-term
experiments with rotations, or experiments that
require frequent monitoring and measurement, are
examples of research that may be quite costly. If
that research seems to offer the best possibility for
resolving a particular production problem, then it
should certainly be considered. But if there are less
costly alternatives, they will probably be given higher
priority. Especially in the first years of on-farm
research in an area, very complex experiments may
distract researchers' attention from establishing a
solid record of collaboration with farmers and
extension agents. As researchers gain more
experience with the area and the difficulties of
managing on-farm experiments, more complicated
experiments can be considered.

In Table 4, several possible solutions present some
questions with respect to ease of experimentation.
The fungicide experiments (6b, c) are rated to be of
medium difficulty because they require large plots.
Incorporating maize residues (1b) and planting
leucaena (1d) get low ratings because they require
several years of experimentation. In the case of
leucaena, researchers believe that going ahead with
investigation is worthwhile despite the fact that a
fairly elaborate experimental program is required. The

incorporation of maize residues was already
eliminated because of incompatibility with the
farming system.

Final Evaluation of Possible Solutions
The last column in Table 4 presents tentative
decisions on the future of each possible solution.
The decisions are tentative because other factors
(such as the importance of each particular problem)
must be considered, as well as the decisions that
will have to be taken later regarding the number and
types of experiments. Nevertheless, it is important
to summarize the analysis that has taken place in
Step 6.

The overall evaluation can be done in several ways.
It is possible to assign scores to each ranking for the
various categories and then add the scores. If a large
number of possible solutions are being considered,
scoring them is often a good way to begin. As only
a small number of possible solutions are being
considered in Table 4, the evaluation was made on a
qualitative basis.

Two solutions included in previous experiments have
performed well enough that work with them will
continue: nitrogen fertilization (la) in maize will be
verified on large plots under farmer management in
preparation for a recommendation, and the new
maize variety (3a) will be part of demonstrations by
the extension agency. The low rating for stability
given to nitrogen fertilization means that researchers
will have to pay particular attention to an analysis of
risk before making a final recommendation. The
evaluation of other solutions (4a, c; 5a, c; 6a, b, and
d) is high enough that they can be considered for the
experimental program. They are all included in List A.
Leucaena (ld) is considered to be worth pursuing as
part of a longer term research effort, and it is
included in List C. The idea of incorporating maize
residues is abandoned until other fodder sources are
developed, a suggestion also included in List C.
Serious doubts about possible solutions 1c, 4b, and
6c cause them to be set aside. Solutions that are to
be included in experiments and require institutional
support are noted in List D.


The sixth and last step in identifying factors for
experimentation is to evaluate solutions proposed in
Step 5. The proposed solutions must be considered
in light of their technical characteristics, the farmers'
ability to adopt the proposed solution, and the
research expense involved. Researchers must
develop a clear set of criteria to evaluate each
proposed solution; seven criteria are suggested here.

The first criterion for evaluating proposed solutions is
the probability that the technology will function
under farmers' agroecological conditions and
management practices. The second criterion is the
estimated profitability of the solution. If either of
these two criteria gets a low rating, the solution
will almost certainly be eliminated from further

The third criterion is whether or not the proposed
solution is compatible with the farming system, i.e.,
with the natural and socioeconomic circumstances
under which farmers operate. The fourth criterion is
the extent to which the solution helps reduce risks
for farmers. The fifth criterion is the need for support
from extension, credit, or input suppliers to ensure
that the solution can be adopted. If researchers have
doubts about any of these criteria then the proposed
solution should be examined very carefully before
work proceeds.

The sixth criterion is the ease with which farmers
can test the proposed solution. The seventh is the
ease of carrying out the experimental program to
test the proposed solution, including the time and
expense required. Neither of these two criteria is
sufficient in itself to eliminate a solution from
consideration, but each is important in deciding
between solutions whose potential is otherwise

Once researchers have rated each proposed solution
on the basis of these criteria, they must come to a
decision regarding the future of the solutions. If a
proposed solution is thought to be acceptable for on-
farm experimentation, it is included in the list of
experimental factors (List A). If a proposed solution
has potential but requires more research before it

can be tested on farms under farmers' conditions, it
is included in the list of themes for longer term
research (List C). If the proposed solution requires
special consideration by extension, credit, or input
suppliers (the fifth criterion above), a note is made in
List D regarding suggested interactions with
appropriate institutions.

Identf.f causes

Analye interrelaiaons
among problems
and causes

Further evidence
required to identify
or evaluate problems

Further evidence
required to determine
causes of problems

UistA Ust B
Factors for Other
experimentztion : diagnostic

Longer term

-strutlo~i -

Summarizing the Six Steps:
Lists of Conclusions

The work described in the six planning steps usually
requires several days of discussion and is based on
many months or years of research. It is therefore
important to summarize conclusions from these
discussions and record them in a form that
researchers can use. One way to summarize
conclusions is through a set of lists. Four lists are
suggested here. List A contains all of the
experimental factors discussed in the six steps. In
particular, it lists the exploratory factors that have
been suggested for examining problems (Step 1) and
causes (Step 3), and the possible solutions to well-
defined problems that received a favorable evaluation
in Step 6. List B contains suggestions for other
diagnostic activities that are useful for obtaining
more information on problems or their causes. List C
summarizes suggestions for longer term research
derived from the evaluation of possible solutions
carried out in Step 6. List D summarizes conclusions
from the evaluation (also done in Step 6) of
institutional support necessary for promoting
adoption of proposed solutions.

List A: Experimental Factors
The principal goal of the planning process is to
develop a list of experimental factors for on-farm
experiments (an example is presented in Table 5).
Those experimental factors will come from
three sources.

1) Exploratory Factors-Some problems (Step 1) or
causes (Step 3) require further experimental
information. The experimental factors will help
researchers explore the importance or the cause
of a problem.

Broad-leaf weed control in beans (using herbicide
Z) is an exploratory factor for better
understanding a possible problem. Phosphorus
placement for maize is an exploratory factor for
examining the cause of phosphorus deficiency in
maize. Exploratory factors do not necessarily
represent possible solutions; a higher than
economic dose of phosphorus may be examined

simply to see if there is a response. But as much
as possible, decisions about which exploratory
factors are appropriate should be guided by the
criteria listed in Table 4 for evaluating solutions.

2) Possible Solutions-If the problem and its
causes) are clear, experimental factors are the
inputs, varieties, or techniques specified as
possible solutions in Step 6. They may have
been tested in previous experiments.

Nitrogen fertilizer application in maize is in the
final stages of testing, and the early maturing
maize variety A is ready for recommendation
(Table 4). Other possible solutions (e.g.,
herbicides C and E or bean varieties J, K, and L)
will be tested for the first time to see if they
give acceptable results. For each possible
solution the group of farmers for whom the
solution is appropriate should be identified.
Small-scale farmers who do not market a high
proportion of their beans will be a target group
for testing bean varieties J, K, and L, but other
farmers will not be interested because those
varieties receive a lower price in the market.
However, all farmers will be interested in
products M and N to control anthracnose. Early
maturing maize variety A is mostly intended for
farmers in the north of the research area.

3) Other Factors-Factors may not come directly
from the identification of problems, causes, or
solutions but should nevertheless be considered
for the experimental program. Such factors are
derived from researchers' knowledge of possible
agronomic interactions with factors being tested.
One example is researchers' belief that rhizobium
should be tested in the presence of potassium.
Another is the fact that experimentation to
control anthracnose should take account of bean
plant density.

It should be emphasized that not all of the factors
that appear in Table 5 may end up in the
experimental program. Their inclusion will depend on
the resources available and the number and types of
experiments that the research team can manage.

Table 5
List A: Summary of Factors for Experimentation

1 Exploratory Factors Sources

Herbicide Z for beans

Phosphorus placement
in maize

To explore whether
broad-leaf weeds are a
problem-Step 1, No. 9

To explore lack of
economic response to
phosphorus-Step 3,
No. 2

80 kg N/ha in split applications
Maize variety A
(particularly for
north of research area)
Herbicide C
Herbicide E
Nitrogen fertilizer
for beans
Rhizobium inoculation
Bean varieties J, K, and L (for
farmers who do not market beans)
Mixture of fungicides M,N
10 new bean lines

Step 6 (Summary of
Table 4)


Bean plant density

Step 6, No. 5c (Note in
column 1, Table 4)
Possible interaction with

Step 6, No. 6 a,b,c,d
(Note in column 1,
Table 4)
Possible interaction with

2 Possible solutions

3 Other factors

List B: Data Needs for Continuing Diagnosis
Some diagnostic activities take place before
experiments begin, but diagnosis should also
continue during experimentation. As researchers
debate the importance of particular problems (Step
1) and seek their causes (Step 3), they may find that
certain diagnostic tools would be helpful and should
be noted. List B (Table 6) gives examples that
summarize the needs for nonexperimentb. data
suggested in Steps 1 and 3. Diagnostic tools may
include a wide range of data-gathering techniques
such as reviews of secondary data, farmer
interviews, field observations, and laboratory tests.
These techniques require additional time from
researchers, and their use should be carefully
planned and integrated as much as possible with the
on-farm experimental program.

Table 6. List B: Data needs for continuing diagnosis
of problems and causes

Data needed


Examine meteorological
data to determine
frequency of drought

Interview farmers who
are beginning to use
herbicides in maize

Field sampling and
laboratory tests to
confirm the species of
organism causing root

Germination tests for
farmers' bean seed

Observation of
emergence rates of
beans under farmers'
soil conditions and land
preparation practices

Step 1, No. 3. Needed
for an analysis of the
problem of drought
stress in maize.

Step 1, No. 4. A simple
informal survey of
herbicide users will give
a better understanding of
the problem of the high
cost of weeding maize.

Step 1, No. 7. To further
describe the problem of
root rots.

Step 3, No. 8. To see if
seed quality is the cause
of low plant populations.

Step 3, No. 8. To see if
soil and tillage conditions
are causes of low plant

List C: Suggestions for Longer Term Research
The analysis carried out during planning is useful not
only for identifying short-term goals for on-farm
experimentation, but also for helping guide longer
term research. As researchers consider possible
solutions to problems in Step 6, they will often
encounter items that may require attention on the
experiment station or through other types of
research. Those items should be noted and discussed
with the appropriate researchers. Examples are given
in Table 7.

Table 7. List C: Suggestions for longer-term research


of fodder



Step 6, Nos. 1b, 5e. Farmers do not
incorporate maize residues because they
are used for fodder; other sources of
fodder appropriate to the region should
be investigated.

Step 6, Nos. 1d, 3b, 5d. Planting
leucaena strips is proposed as a possible
solution for problems related to soil
fertility and moisture retention; initial
research needs to be done to identify
appropriate varieties of leucaena,
investigate how they might be planted,
and analyze their economic viability.

List D: Suggestions for
Improving Institutional Support
The evaluation of possible solutions in Step 6 may
produce suggestions for improving institutional
support. A review of column 5 in Table 4 should
help summarize issues to be discussed with
extension agents or policymakers (Table 8).

Extension agents should be involved in all stages of
OFR and should assume responsibility for much of
the work in verifying and demonstrating new
technologies. In addition, they should be particularly
involved in experiments with possible solutions that
will require special extension programs, which can be
noted in List D.

The availability of inputs cannot be overlooked. If
new varieties are recommended, researchers should
make sure that a seed supply system is in place.
Researchers are responsible for discussing with both
public and private input suppliers the implications of
research results for input availability and quality. It
should be emphasized that proposals to change the
supply of inputs may take considerable preparation
as well as a study of the rationale for current policy.
It should not be assumed that simply informing
policymakers of the advantages of new inputs will
be sufficient to bring about change.

Data from on-farm experiments may also be used to
suggest changes in the composition or requirements
of a credit program, as is the case with fertilizer in
the example.

Table 8.
List D: Improving institutional support


Extension agents should be included in the testing
of nitrogen fertilizer in maize (la), pre-emergence
herbicides (4a, c), and rhizobium inoculation (5c),
so that they can begin to consider possible
extension strategies for those technologies.

Researchers need to be sure that a seed
production system is in place for maize variety A,
which is being demonstrated to farmers (3a), and
they must also investigate the possibility of seed
production for bean varieties J, K, and L (6a). In
addition, researchers must talk to input
distributors to examine whether supplies of
herbicide E (4c) can be assured.


Credit officials should be made aware of the
experiments with fertilizer on maize (la) and
beans (5a). If those solutions prove to be
acceptable, it would be helpful if more credit were


The conclusions of the planning steps are
summarized in four lists. The first contains all the
factors that are to be considered for on-farm
experiments, including exploratory factors that help
develop information on problems and their causes, as
well as possible solutions to problems that are well-
defined. The second list includes suggestions for
further diagnostic activities related to the problems
or their causes. The third list summarizes proposals
for longer term research related to the possible
solutions, and the fourth contains suggestions for
ensuring that institutional support is available for the
solutions under discussion.


Further evidence
required to identify
Sor evaluate problems
Identify causes ;

Analyze inlerrelatlons
Among problems
and pauses
Sand causes
turner ev.aence
required to determine
causes of problems

UK: a.

lPaun rMftor 1 I Other

Longer term

support '

A Final Word

on Priorities

The steps presented in this document are a guide to
setting priorities for on-farm experimentation. The
results of that analysis are summarized in four lists,
and the items listed may be subject to further
scrutiny. Even after much debate and the elimination
of many research themes suggested during the
planning session, the final lists may still contain more
items than the research program can manage, and
further decisions will have to be made. Planning to
do more research than the resources and personnel
of the program can accommodate results in half-
done studies, unanalyzed or even unharvested
experiments, and much wasted effort.

The experimental factors in Summary List A have
already been subjected to two reviews in which
decisions were made about the importance of the
problems and their interrelations. The experimental
factors will be subjected to further analysis when
decisions are taken about experimental design: how
many types of experiments can be managed and
how many factors should be placed in one
experiment?14 The items in Summary Lists B, C, and
D should also be examined carefully. Which
diagnostic activities are most important and can be
managed by the research team? What are the
priorities for longer-term research? Which issues
merit the special analysis and preparation that
discussions with policymakers require?

Those are difficult decisions. The steps presented in
this document are only a guide to setting priorities.
They are not formulas for decision-making, but rather
suggestions for managing the debate, and their real
utility depends very much on the energy and
imagination of the researchers who use them.

14 Exactly which factors can be accommodated in an
experimental program depends on which of them fit together
in efficiently designed experiments, as well as on the
resources available for the experimental program. See
Woolley (1987).


Byerlee, D., M. Collinson, et al. 1980. Planning
Technologies Appropriate for Farmers: Concepts
and Procedures. Mexico, D.F.: CIMMYT.

CIMMYT. 1988. From Agronomic Data to Farmer
Recommendations: An Economics Training
Manual. Completely Revised Edition. Mexico, D.F.:

Delp, P., A. Thesen, J. Motiwalla, and N. Seshadri.
1977. Systems Tools for Project Planning.
Bloomington, Indiana: PASITAM.

Gomez, K.A., and A.A. Gomez. 1984. Statistical
Procedures for Agricultural Research. Second
edition. New York: John Wiley.

Harrington, L., and R. Tripp. 1984. Recommendation
Domains: A Framework for On-Farm Research.
International Maize and Wheat Improvement
Center Economics Working Paper 02/84. Mexico,

Huxley, P.A., and P.J. Wood. n.d. Technology and
Research Considerations in ICRAF's "Diagnosis
and Design" Procedures. International Council for
Research in Agro-forestry Working Paper No.26.
Nairobi, Kenya: ICRAF.

Mutsaers, H.J.W. 1985. An Approach to the
Organization of On-Farm Research Training
Workshops. International Institute of Tropical
Agriculture OFR Bulletin No. 3. Ibadan, Nigeria:

Rhoades, R.E., and R.H. Booth. 1982. Farmer-Back-
To-Farmer: A Model for Generating Acceptable
Agricultural Technology. International Potato
Center Social Science Department Working Paper
1982-1. Lima, Peru: CIP.

Van Der Veen, M.G. 1984. Setting research
priorities: A review. International Rice Research
Institute Training Module D 06. Los Baios,
Philippines: IRRI.

Woolley, J. 1987. The Design of Experiments for
On-Farm Research. Draft Working Document. Cali,
Colombia: CIAT.

Design: Anita Albert (Iowa State University)
Editing: Kelly Cassaday (CIMMYT)
Layout and Production: Miguel Mellado, Jos6
Manuel Fouilloux, Rafael de la Colina and
Bertha Regalado (CIMMYT)
Typesetting: Maricela A. de Ramos and
Ma. Teresa Pedroza (CIMMYT)

o P q
t t


%, "I V, Q 1, i 'o,
4'i Wow k,
o PV


I, it


lal tn
W -4 Pvc OR 4
N -OA xim"


t q

rX on

!j", o
til a+a?
V -nxtt', jil 54 VQ; %1-0 '1
`oi A as -11,
a, l", iI, olo""'t . . .
lk n
0--tt, I r A n

A m nt,
X_ Mao
'A ppowt

4 %ori v"",
o A
M_ Ow, 5 o
o"oJ-4&4q I
A&M 2"Al
'4't 4,


ply, r



An, -I Io,
N, wo

I[q too N Q
4 [n,
Mch I l W,
n, -,l"p oil

7 K

a l . . ........
i'l tot Alto,

No 3 fp f
T Most;

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs